Transforming EDC-The Emerging eR&D Model

February 2, 2004
Applied Clinical Trials
Volume 0, Issue 0

As so many articles have noted,1 the uptake of electronic data capture (EDC) and associated technologies has continued to be disappointing. Even though 20 years ago experts predicted that electronic case report forms (CRFs) would replace paper as a natural consequence of the introduction of computers, this has not been the case.

As so many articles have noted,1 the uptake of electronic data capture (EDC) and associated technologies has continued to be disappointing. Even though 20 years ago experts predicted that electronic case report forms (CRFs) would replace paper as a natural consequence of the introduction of computers, this has not been the case.

Recent analyst research2 has identified that at best only 30% of ongoing clinical trials are conducted using EDC. Justification for this slow acceptance has come from many varied sources. Before we consider this, however, let's examine what a pharmaceutical company looks for when selecting a software solution vendor.

Financial and organizational stability. A pharmaceutical company will not commit to a vendor unless it has a solid balance sheet, funds to resource its current trial support, and proven long-term sustainability as a business.

Global presence and scalability. Although there are some requirements for local providers, the stage for the majority of clinical trials is global; a vendor company needs the infrastructure to support this.

Experience in the marketplace. This is a challenging focus area, where failure is not an option. For this reason, vendors that are viewed as "e-veterans" are often favored.

Robust, reliable technology. This should be encapsulated by excellent support processes.

A cursory glance at the eClinical vendor market will show as many as 150 companies. Although many cannot offer the full package, an emerging subset of vendors do possess these desired characteristics. So perhaps this slow adoption rate of e-technologies cannot be fully attributed to the so-called "unstable vendor market."2

In addition to the desired vendor characteristics, the pharmaceutical company needs to consider the demands of the clinical program.

  • Will the external stakeholders accept the technology? Do they actually want it?

  • Is the infrastructure stable enough to conduct EDC trials in all of the countries which have the desired patient populations (often Central European or African/Asian countries)?

  • How is source data verification managed in this environment?

  • How will EDC affect the internal organization?

  • What is the opinion of the regulatory authorities of data collected in this way?

  • Is there a danger of data being rejected as noncompliant?

  • When will we see a return on investment? How can this be measured when so many of the EDC benefits are qualitative rather than quantitative?

In their efforts to address these challenges, the approach of some pharmaceutical companies has been laudable. Their piloting trials have taught them the issues are more about people and process than about technology.3 Figure 1 illustrates the challenges confronting pharmaceutical companies that employ EDC. Instead of extended use of the technology, various issues have led to consecutive piloting.

Current "EDC double data entry" paradigm

Pharmaceutical companies have recognized that going eClinical changes the trial process from a series of sequential tasks to a parallel process where a continual review of data will bring about the best results. Impact analyses were conducted to examine the change that EDC would bring, and these resulted in some restructuring efforts within the organizations. There have been investments in developing the functionality of the EDC system so that it is not only an application to capture clinical data, but also for review and analysis. But even with all of this work, the growth in acceptance of EDC has not been exponential as predicted. How can this be explained?

Figure 1. Expected electronic data capture industry growth has turned into consecutive piloting.

If we look at the most critical aspect of the clinical trial-the investigator and the site team-we can see that EDC has not been the correct solution. Although we have made improvements to system functionality, invested in high-speed connections to the Internet, and even provided extra resources for data entry, the bottom line is that EDC is an additional burden to the site. For the majority of clinical trials, due to the uncertainty surrounding the regulations, we have doubled the work for the sites.

In the paper environment, the paper CRF was received in-house and entered by one data entry clerk and verified by another. This double data entry process was supposed to be removed by the introduction of EDC. The reality is that for the sponsor organization it has disappeared, because the sites now perform this redundancy.

This "EDC double data entry" is caused by the assumption that a paper source is needed for online EDC systems. That means, essentially, that the investigator records the patient data in the source notes or a pro forma of some kind, and then enters it into the EDC system some time later. In a recent survey conducted by the industry group EDM Forum,4 840 investigators (both experienced and nae in EDC) were asked what the biggest barrier was to wide-scale use of EDC: this double data entry process was number one. Until global health care evolves to the point where all patient data is stored electronically and we are in a position to copy selected views of this information in a validated exchange, we need a solution to reduce the site burden of EDC and clinical trial conduct in general.

It is evident that the process reengineering and impact analysis conducted by the pharmaceutical organizations do not go far enough. The clinical trial process was conceived, designed, and refined using paper as the medium for collecting clinical data. All the procedures and functional roles that make up the clinical data process have been structured around paper. Although there have been some efforts to recognize this internally, the external processes and the site expectations have not been modified.

As an Industry, we need to ask ourselves two simple questions:

1: What data do we want to collect for our clinical trials?

Naturally, the protocol and therapeutic area under study primarily drive this. But typical clinical data can be roughly divided into two groups-physiological data such as vital sign readings, laboratory values, and medical history; and patient-reported outcome (PRO) data such as symptom counts, subjective pain analysis, and quality of life assessments. These two categories historically make up the typical clinical trial,5 with physiological data accounting for 75% and PRO 25%.

Over the past few months there has been a groundswell of interest in PRO data. PRO data is now being used to assess primary and secondary endpoints of clinical trials, and the FDA has acknowledged that PRO data makes up an important part of a NDA.6 This regulatory push, combined with the current "in vogue" therapeutic areas under investigation (sexual dysfunction, CNS in general, asthma, osteoarthritis), has seen the relative proportion of PRO data increase in a clinical trial.

Figure 2. The eR&D model modular overview.

2: At what point and from whom do we want to collect it?

In a clinical trial, the interaction between the investigator and patient is limited to the visit, intervals between which can vary from weeks to months. CRFs include questions such as "How have you felt since the last visit?" or "Have you experienced any pain since the last visit?" or "Have you taken your medicine at the correct time?"

As we all understand, it is difficult to recall this information over long periods of time, so the patient is asked to record this information in a paper-based patient diary they use at home. The concept is that the information recorded at the medical moment will therefore increase its accuracy.

There is legitimate concern that while PRO data clearly contributes important information to the NDA, paper diaries produce data that is notoriously poor7 (as anyone with a background in clinical trials will tell you). Patients forget to complete the diary at the correct time, and often back- or forward-fill the diary. Considering that this method is being used to assess primary and secondary endpoints, it's a serious concern that the majority of this data is at best irrelevant and at worst fraudulent.

Electronic Patient Diaries (EPDs)

Electronic Patient Diaries have become increasing popular where PRO data is critical to the success of the clinical trial. Instead of using a paper diary, the patient enters data directly into a mobile device. The EPD actively reminds patients to be protocol-compliant regarding when they fill in the diary, instead of filling the diary data just before seeing the investigator.

But what, you may well be asking, do electronic patient diaries have to do with reducing the site burden of EDC? In the following case study, we demonstrate that the intelligent introduction of technology, alongside the correct user populations, can engage and encourage the sites to embrace eClinical technologies.

Case Study-Combination of EDC and EPD to yield maximum success8

The study in question was a Phase III IBS trial with 5,000 patients in 23 countries. It is anticipated that over 2,000,000 pages of diary data will be processed for this study. The goals of the study team were to:

  • Collect primary efficacy data reliably, direct from the patient

  • Improve the quality of the raw data collected from the patient

  • Provide a view into the patients' home life to assess how well the condition can be managed

  • Keep clinical development time and costs to a minimum while maintaining data quality.

All patients were provided with electronic patient diaries, and were requested to complete the diary when the device prompted them to do so. After each diary was completed, the data was transmitted to a secure central database. This data was then presented to the sponsor and site staff in a series of reports, viewable through a secure, Web-based review tool.

Key benefits currently realized

This trial is currently ongoing, but the study team has already seen the following benefits.

Improved compliance. The ongoing level of compliance is 97%. That is to say, compared to the protocol only 3% of the data expected is missing. Not only is this a tremendous amount of compliant and accurate data, it has all been collected within the appropriate timeframe.

Screening assessment using EPDs. In a bid to recruit the correct patients the first time and reduce dropout, the EPD was used to screen patients. At the screening visit following the preliminary assessment by the investigator, each patient was given an EPD and asked to complete the diary during his or her daily routine. During the screening period the data is transmitted to the central database and reviewed by the patient. The Web-based reviewer tool runs a series of algorithms against the data to calculate the eligibility of the patient. At the next visit, the investigator can then move eligible patients to the next phase of the study, or retrieve the EPD from ineligible patients and enroll another round. This dense data sampling is anticipated to reduce the required number of patients to be randomized by 500, which translates into a two-month time benefit.

Online protocol amendments. The EPD technology supports the online protocol amendments, which can be sent to the patients' EPDs to allow flexible rule adjustments. It was noticed at the beginning of the study that the level of screening failures was higher than expected. On closer examination, one of the inclusion criteria was inappropriate in this patient population. An amendment was made, and the level of randomized patients increased. This meant that the trial could be proactively, rather than reactively, managed.

Reduced site administration. The trial consisted of several complex phase and dosing transitions, which would have normally placed an additional burden on the site for the duration of the trial. With the intelligent use of EPDs and rules running across the Web-based reviewer tool, the site could use the technology to help administer the patients and their treatment.

Cost-neutral operation. Once a patient had completed the study, the EPD could be recycled and given to the next patient recruited. There was no need for the EPD to be returned to the vendor for configuration, since this was all managed on-site. As a direct result, the investment required to purchase the devices can be amortized across the study to create a cost-neutral operation.

In summary, the EPDs were used to collect critical PRO data directly from the patient. The patients were allowed to set when the alarms and diary reminders were triggered, giving them the ability to fit the diaries into their lifestyle. The data transmitted from each patient was sent to our secure server and made available via a series of summary reports presented via a Web-based reviewer tool to the investigator and the site team to aid the site in patient management and administration. Although the sites needed to manage the EPDs and keep the batteries charged, they felt that the technology reduced the overall levels of site administration and workload for the trial.

The physiological data was captured using an EDC system. As a consequence of the intelligent use of EPDs for collecting a large proportion of the data, the actual amount of data entered into the EDC system was significantly reduced; therefore the EDC burden, although not removed, was certainly diluted.

For the first time, the external end-users of eClinical technologies could see the tangible benefits that the pharmaceutical industry has promised them. This trial will act as a catalyst for change in the way we collect data and conduct clinical trials. In short, it is the emergence of the e-Research & Development (eR&D) model.

The eR&D model

As we have seen in the case study, refocusing the primary data capture point from the site to the patient results in some favorable changes to the existing workflow in clinical trials. For the purpose of brevity in this article, let us divide the eR&D model between the site and the sponsor.

Site-As we have previously discussed, the patients enter data in their EPDs, and this data is transmitted back to a secure central server. We acknowledge that not all needed data can be captured in this way, so the patients visit the site periodically as per protocol. We do believe, however, that the visit schedule could be reduced in this model, but naturally this would depend on the therapeutic area.

As the data is securely transmitted to the server, the investigator can review each patient's data online throughout the conduct of the trial. So, rather than reducing the relationship between the investigator and the patient, eR&D actually strengthens it. As in the case study, this Web-based reviewer tool could also be potentially used to monitor primary end points and as a consequence used to modify the medications that the patient is taking. This proactive and personalized patient care is only possible with EPDs.

For data that cannot be captured in this way, the EDC system is used at the site. Once the data has been entered, this should ideally be presented together with the EPD-derived data through a single reviewer tool as the complete patient profile. The eR&D model promotes a modular approach to capturing clinical data where the clinical team can determine what are the most appropriate methods for data capture according to the patient population (see Figure 2). This creates an eClinical toolbox, where the clinical team selects the correct tools for the specific trial and the data integrates regardless of tool selection.

Sponsor Clinical research associates (CRAs)

As the data is being collected, the CRA is able to review it online and in real time. For the diary data, there is only one source, which is stored at the database level. Diary devices can be viewed as transient data collectors that acquire data, store it in files temporarily, and as part of normal workflow pass it into databases before the process task is complete. The net result of this is that source data verification is not required as there are no copies of the source to confirm. For EDC data, however, there may well be source copies to confirm, and this can be performed according to the protocol. The amount of verification is reduced through this model and, when the data becomes available online, any issues with the trial can be identified before the site visit and appropriate measures put into place before it affects the conduct of the trial.

The key advantages for the CRA, however, are within the Web-based application. They now have the information available to evaluate the site's productivity and therefore control recruitment so that targets can be managed much more efficiently. The reviewer tool will reveal which patients are being noncompliant, so CRAs can drop them from the study and recruit new patients. In the paper environment, proactive trial management was not possible, and so recruitment and site management was as much about estimation as it was about information.

Extending this further, the CRAs can then control the drug supply to each site, reducing unnecessary delays or waste. On this basis they can control the payment to the sites to ensure that each site provides a good service. Since this service data is online, the investigator can also track his payments based on the same information.

This model allows the CRAs to perform the job that they are required to do more effectively and will facilitate in the smooth running of the trial.

Clinical data management (CDM)

EPD data is collected from the patient as verbatim and is therefore not subject to investigator amendment. To leverage the advantages of the EPD, edit checks are programmed into the device to ensure the accuracy of the recorded data. The Web-based reviewer tool will allow CDM officials, along with the CRA and the investigator, to monitor patient compliance with regard to diary data completion.

For the EDC component of the trial, as we are collecting less data, there should also be a corresponding reduction in the number of queries raised-benefiting the site and the sponsor. When developing the edit check specification, care should be taken to ensure that no "noise"-that is, no unnecessary queries-gets raised. There should also be a clear distinction between the checks that are immediately raised (for example: "The height if the patient is out of range, please qualify") and those multiform, multivisit checks that require the skilled review of a data manager before being issued to the investigator.

As far as possible, all edit checks should be performed in the front-end EDC system, rather than the back-end data repository or clinical data management system (CDMS). This will avoid any discontinuous query generation and will ensure that queries are resolved in stream and within the revenue-generating period for the investigator.

External data, such as labs, can either be loaded into the EDC system or transferred to the back-end CDMS/repository structure (if available). Before making this choice, it is necessary to consider what value is added by loading this data into the EDC system and whether it is necessary for the investigator to review the data.

Coding of medical terms also has to be performed, a task probably best handled through the back-end CDMS/repository. Data from the EDC/EPD system can be routinely exported here and coded. Any queries resulting from this can be pushed through the EDC system. If the organization does not possess a coding tool, then it may be possible to use a Web-based coding application and maintain all clinical data in this enterprise model.

The traditional methods of data management are no longer appropriate in the eR&D model. That is not to say that it is a redundant function. Rather, data managers can focus on the overall trial data quality, rather than struggling with basic data cleaning, since the technology can deal with this. As a direct result, the productivity of the data managers increases, as they are able to manage more clinical trials.

Pharmacovigilance

The requirement of the pharmacovigilance department in this model is simple: immediate notification of serious adverse events (SAEs). The real-time nature of the data lends itself extremely well to this, and most EDC systems can send automatic notification (email, SMS, wireless alerts) to the appropriate personnel when a SAE is recorded. The additional data required for SAE reporting should be captured in the EDC tool as well. No EDC system would or should possess the functional characteristics of an adverse event reporting system, and so the captured data should then be transferred to such a system for adverse event report generation according to the regional regulations.

In the future there may be the scope to use EPDs for self-reporting of adverse events which, following analysis by qualified pharmacovigilance staff, would result in direct interrogation of the patient to establish whether the event was serious or not.

Conclusion

In conclusion, the proposed eR&D model represents considerable benefits for all stakeholders involved in the clinical trial. Although we are at the beginning of this evolution, the case study described demonstrates immediate and tangible benefits.

For the patient, the primary data capture point is focused upon them. Rather than being a burden, this allows flexibility in the way their data is captured and, in the future, may lead to tailored treatment.

For the site, the level of EDC data entry is significantly reduced, and the EPD data is available online. This adds value by allowing the site personnel to fulfil their main objective-individualized and improved patient care. As was seen in the case study earlier in this article, sites can see and feel the benefits of using this technology. As with the adoption and acceptance of any new technology, if the end-users can see the benefits in their daily routine, the level of adoption will increase as a natural consequence.

For the sponsor there are many benefits. Use of EPD and EDC delivers quality data faster. As the eR&D model brings the patient in to the center of the trial, thus creating a window into the patients' lives, it allows proactive intervention and trial management based on the real-time access to data. This more effective trial management has demonstrated significant financial benefits by decreasing dropout rates, speeding trial completion, and allowing data-driven decision-making at program, trial, site and patient levels. This has already been seen with EDC initiatives, but what the eR&D workflow delivers in addition is the potential to reduce the visit schedules of the clinical trials without losing control of the patients. This will automatically lead to a reduction in the number of monitoring visits, but will actually increase the level of proactive trial management via the Web-based reviewer tool. CDM workload will be reduced and productivity will increase with no decline in data quality.

Naturally, we recognize that not all trials can be conducted in this way, but we do believe that the proposed eR&D model could be adapted for many settings. It is always difficult to balance the needs of the clinic with the needs of the clinical trial, and naturally the clinical trial will always be secondary to the care of the patient, but with the eR&D model all stakeholders can meet their objectives.

References

1. M. Miller Smith, "EDC Use in Clinical Trials: Are We on the Verge of a Breakthrough?" EPC, December 2002.

2. CenterWatch-CDISC Survey, November 2002.

3. L. Stubbs, "Using EDC in Phase 1-The J&J experience," ACT Euro Summit 2003.

4. Research Survey, Electronic Data Management Forum, September 2002.

5. Unpublished material, DataEdge, 1999.

6. J. Weschler, "Patient Reported Outcomes Gain Credibility," Pharmaceutical Executive, May 2003.

7. See for example, A. Stone, S. Shiffman, J.E. Schwartz, J.E. Broderick, M.R. Hufford, "Patient Non-compliance with Paper Diaries," BMJ, 324, 1193-1194 (2002), and C.B. Johannes, S.L. Crawford, J. Woods, R.B. Goldstein, D. Tran, S. Mehrotra, K.B. Johnson, N. Santoro, "An Electronic Menstrual Cycle Calendar: Comparison of Data Quality with a Paper Version," Menopause, 7, 200-208 (2000).

8. J. Phillips, "The Economic Use of Direct from Patient Data Capture technology," DIA Europe, March 2003.

Tim Davis is director of corporate development, CRF Inc. (formerly CRF Box), Tammasaarenkatu 5, FIN-00180 Helsinki, Finland, email: tim.davis@crfhealth.com, www.crfhealth.com.