Validating electronic source data in clinical trials

Transcription

1 Controlled Clinical Trials 25 (2004) Validating electronic source data in clinical trials Ronald G. Marks* Department of Biostatistics, College of Medicine, University of Florida, P.O. Box , Gainesville, FL , United States Received 5 March 2004; accepted 6 July 2004 Abstract The clinical trials industry relies heavily on paper-based source documents as the foundation for the collection of its clinical research data from human subjects and medical records. This focus on paper documents has been prevalent throughout the history of clinical trials conduct, even as computing solutions advanced throughout the past 20 years. With the advent of additional electronic capabilities recently with the growth of Internet-based products to enhance business operations in many fields, the clinical trials industry remains uniquely behind most other industries in electronic technology adoptions. Valid reasons exist for the slow growth of technology adoptions in clinical trial activities, but there are now discussions about how to use technology more effectively in clinical trial conduct. One area of enhanced clinical trial conduct is believed to be available by moving from paperbased source documents to electronic source documents, that is, eliminating paper from clinical data capture, and collecting the information initially in a computer system. An important concern in moving to electronic source data is the validation of such data. This paper summarizes the history of clinical data capture through paper and electronic advancements to date and identifies three reasons for the slow movement to more electronic source data. The paper then illustrates two methods for the validation of electronic source data. D 2004 Elsevier Inc. All rights reserved. Keywords: Internet-based products; Paper-based source documents; Electronic source documents * Tel.: ; fax: address: ronmarks@ufl.edu /$ - see front matter D 2004 Elsevier Inc. All rights reserved. doi: /j.cct

2 438 R.G. Marks / Controlled Clinical Trials 25 (2004) Introduction The scientific conduct of clinical trials is critical to the evaluation of newly developed products in the pharmaceutical, biotechnology, and medical device companies. One major area of conduct in a clinical trial is the collection of clinical data from study subjects by physicians, nurses, physician assistants (PA), and other appropriately trained medical personnel. Historically, this clinical data has been first recorded on paper by the medical professional, and then computerized for analysis. This process is quite lengthy, cumbersome, and prone to error that requires substantial human intervention to complete. FDA requirements to ensure resulting data is as accurate and complete as possible complicate the process. This clinical data capture process has evolved over time in the industry. In the late 1990s, it was believed that the introduction of web-based technology provided an opportunity to greatly improve the efficiency and accuracy of clinical data capture. Electronic data capture (EDC) systems became available in the marketplace with the expectation that efficiencies gained in other web-based markets would now be brought to clinical data capture. To date, many would agree that such efficiencies are still not apparent, mainly due to the continued use of processes involving paper-based data collection. The industry is now beginning to discuss the opportunities for true electronic data capture by replacing dpaper-based source datat with delectronic source datat. This paper will summarize the historical development of clinical data processes and illustrate why major inefficiencies still exist in the process in spite of the evolution into web-based systems, mainly due to the continued reliance on paper. A main hindrance in moving to more electronic source data is the belief that electronic data cannot be validated a requirement of the FDA. This paper will illustrate two validation methods for electronic source data. An additional benefit of using electronic source data is that the medical professionals collecting the data are treated like medical professionals instead of the ddata entry clerkt that is the current perception. Major improvements are needed in the capture of clinical data in clinical trials to improve the cost and timeliness of this essential component of clinical trials. These gains have not yet been realized in spite of the introduction of web-based EDC tools. Impressive gains in data collection efficiency, accuracy, subject safety, and treatment of medical professionals are available with the implementation of true electronic source data collection methods. 2. Evolution of clinical data capture In the early 1980s, personal computers (PCs) were introduced and soon became commonly used tools for business and personal tasks. By the mid-1980s, PCs were introduced to clinical trial use for clinical data capture. Use of PCs for this purpose led to a major transformation in the way clinical data was captured. Before PC use for clinical data capture, site professionals captured data on paper case report forms (CRF) and sent the forms to a sponsor centralized facility where data computerization took place. This method of data capture was called dcentralizedt because the data was computerized in a single facility by professional data entry personnel. The investigator s main responsibility was the original completion of the paper CRFs, and then responding to queries that arose from the sponsor after reviewing the computerized data.

3 R.G. Marks / Controlled Clinical Trials 25 (2004) PCs at the investigator site allowed for the introduction of ddecentralizedt clinical data capture which became known as remote data capture (RDE). In this approach, the investigator and staff still completed paper CRFs. However, a major change was that the staff member would now also computerize the clinical data. The reasoning was that the centralized system led to long times from original data capture to computerization to data validation. Common errors were of data fields not completed, or completed in error, or using illegible writing. Long times between original data capture on paper and validation made validation quite burdensome. Sponsors believed that site computerization of the forms should reduce each of the types of errors made and reduce the time from initial data capture to computerization to data checking by the sponsor. It was believed errors would be reduced as staff computerizing data shortly after collecting it would catch and correct many errors that would otherwise be missed. This led to a major paradigm shift in clinical trial conduct placing the workload and responsibility for data computerization on site personnel. Sponsors developed proprietary hardware and software solutions to manage their RDE at investigator sites. These systems allowed for some edit checking of data as it was entered, which was believed would reduce the number of errors getting into the data in its initial computerization. If a staff member entering data had a questionable data element, they had the subject CRF handy to check the result. Over time, RDE matured and became more efficient as site investigators and sponsors became more proficient in this new way of collecting clinical research data. In RDE, computerized data was routinely transferred from each investigator site to the sponsor through some File Transfer Process (FTP). This process might be done daily, weekly, or using whatever schedule worked best for an individual study and investigator. The FTP process was usually done via phone and might take quite some time to complete depending on the volume of data to transfer. Then, in the late 1990s, web-based approaches to clinical data capture were introduced. It was believed that efficiencies would be gained as had been achieved and documented in other industries moving processes to the Internet. The acronym was also now changed from RDE to EDC electronic data capture. In some fundamental ways, EDC was not different from RDE. Site investigators still completed paper forms and still computerized the data for transfer to the sponsor. Data transfer was now expedited by the Internet instead of by FTP, resulting in more frequent and rapid data transfers. It was also expected that more data checks could be built into CRFs, reducing errors upfront and efficiently reducing the query rates. An additional benefit would be to replace proprietary hardware and software systems required by sponsors for their RDE systems more open systems could now be used taking advantage of Internet software. The reviews of EDC are mixed to date, but few believe EDC has provided the efficiency gains expected when it began. One of the important reasons for this lack of success is the lack of commitment to process change required to achieve the sizable benefits industry believes are possible. Process change requires looking at the complete process and improving it, not simply continuing an inefficient process in a web-based approach. The reality is that the process of collecting clinical data is essentially the same as it was through the 1970s when data was computerized dcentrallyt and continuing through the 1980s and 1990s using RDE and now using EDC. Efficiency gains are possible in EDC when we look to identify the inefficient processes in current EDC and work to improve them in conjunction with the use of the Internet.

4 440 R.G. Marks / Controlled Clinical Trials 25 (2004) EDC issues The greatest process inefficiency in EDC revolves around the paper-based source documents commonly used in most industry clinical trials. Whereas most other industries have achieved efficiency gains through web-based systems, the gains have usually been achieved by also changing their processes from paper-based to electronic. Two notable examples are the banking and airline industries. The banking industry regularly moves large amounts of money electronically around the world in a secure and timely manner. The airline industry manages millions of passenger reservations globally electronically in a secure, safe, and reliable manner. In both industries, paper copies of transactions are available as requested but not required. Yet the clinical trials industry has been reluctant to make this important and essential process change to improve the clinical trials conduct. There are at least three reasons for this lack of motivation to change: (1) FDA, (2) lack of mobile hardware, and (3) attitude. The FDA requires validation of all clinical data from each trial and provides guidance requirements for electronic systems capturing clinical data [1]. Although these guidelines are developed and discussed for the use of belectronic systemsq, much confusion has existed over these guidelines due to the use of paper-based source documents being used in most clinical trials. The confusion has led to the FDA reexamining those guidelines with a plan to provide an updated interpretation [2]. A second issue hindering progress in electronic research data is the lack of portable hardware for data gathering. Until recently, investigators had to use PCs hard wired to walls and telephone modems that often were too slow and inaccessible for efficient electronic data collection. With the advent of highspeed wireless networks and lightweight portable computers, this issue should begin to become less so in the future, especially in physician offices. In addition, physicians are replacing paper-based medical record systems with electronic medical record systems (EMR), so this will increase their comfort and use of electronic research systems as well. Hospital data collection is more problematic due to firewall and bureaucratic issues of capturing clinical research data, but opportunities are also improving here. A third issue impacting progress is the attitude of monitors whose job involves the validation of the collected clinical data. The manual validation processes that have been in place since the centralized data computerization through RDE and now EDC are so deeply ingrained that it is difficult for many monitors to consider any process that does not include paper. Monitors need to realize that validation will still be required for electronic source data, but their processes will also change, allowing them to become more efficient in their processes as well. 4. Validating electronic source data Clinical data comes from three main sources as illustrated in Fig. 1 (1) medical record, (2) directly from the subject, and (3) lab tests. Note that electronic lab data has been accepted by the FDA for many years, as illustrated in Fig. 2. Laboratory data capture has been routinely conducted in an electronic process and source data is considered to be the electronic data files provided by the lab to the sponsor.

5 R.G. Marks / Controlled Clinical Trials 25 (2004) Fig. 1. The three core sources of clinical research data: (1) medical record, (2) directly from the subject and (3) laboratory test. Hence, there is certainly a precedent for electronic source data in the clinical trial arena. The current validation issue is considered for the area of clinical data that is either captured (1) from the subject directly or (2) from the subject s medical records. In both these instances, currently, paper is generally considered the source for which validation must be conducted. A solution is proposed here that allows for both these data types to be considered as electronic source data, with appropriate validation methods. Fig. 2. Most clinical research lab data is electronically entered directly into a computer, with the electronic data file serving as the source data.

6 442 R.G. Marks / Controlled Clinical Trials 25 (2004) These two areas will be considered separately for validation purposes, but it will be shown that the validation procedure is similar Validating direct data capture from the subject Fig. 3 illustrates the two ways of collecting data directly from a subject: (a) a two-step approach (1) first recording the information onto a paper CRF and then (2) computerizing from the paper source document as is most common today, or (b) a single-step approach of direct entry onto an electronic CRF producing an electronic source document. The major concern with data collected directly from the subject into an electronic format is the perception that there is nothing to validate. If data is first written on paper, the paper CRF becomes the dofficialt data that is accepted as correct throughout the study whether or not errors were made in transcribing the data onto the CRF. Without the paper, how can we ever be sure the electronic data is correct? Fig. 4 illustrates inclusion/exclusion (or clinical) data (A) entered directly onto an electronic screen as the medical professional captures it from a subject. Upon hitting the dsubmitt button, a second screen appears (B) summarizing the data just entered. The medical professional reviews the entered data a second time to insure its completeness and correctness. If there is any reason to correct any data elements, hitting the browser dbackt button returns the medical professional to the previous Fig. 3. Clinical research data captured directly from the subject can be entered in two steps via (1) paper case report form to (2) computerization, or in one step (1a) directly into the computer.

7 R.G. Marks / Controlled Clinical Trials 25 (2004) Fig. 4. Clinical research data entered (A) directly into the computer can be validated by having a second screen appear (B) requiring the investigator to validate the data at the point of entry. electronic CRF to correct any entries. When the professional is comfortable the data is correct, hitting dsubmitt sends the data to the database. The medical professional has performed this work using their unique username and password. In addition, the medical investigator has signed a document acknowledging that all work performed under this username and password is their responsibility, thereby the username and password is serving as their electronic signature. This electronic data gathering approach closely mirrors regular clinical medical data capture, and allows validation by the medical professional that best knows the subject and the data elements being captured and with the subject still present. This approach greatly reduces the need for monitors to validate all data points, and reduces the monitoring burden on the site Validating data capture from a subject medical record A common misconception is that research data taken from a patient medical record is automatically paper-source as it is already existing on the paper medical record. Whether copied onto a paper CRF first or directly computerized from the medical record as illustrated in Fig. 5, many clinical trial professionals

8 444 R.G. Marks / Controlled Clinical Trials 25 (2004) Fig. 5. Clinical research data captured from a medical record can be entered in two steps via (1) paper case report form to (2) computerization, or in one step (1a) directly into the computer. believe the medical record becomes the source document, and computerized data must be compared to the medical record. Instead, if the same system is used for capturing research information from a medical record as described above for the data captured directly from the subject, the research data entered electronically from the paper medical record and validated immediately by the medical professional would constitute electronic source research data that was already validated. In some instances, the medical professional might not wish to review each CRF upon completion as suggested above. A second approach to validation would provide a single text report of all recorded information across multiple CRFs to review at one time. This approach could be used instead of validating each CRF as illustrated in Fig. 4. Fig. 6 illustrates the approach to providing a complete text summary of all entered information. In this approach, the investigator reviews all entered research data at one time in a text-based format. If any data elements are questionable, the investigator can return to the appropriate CRF to clarify that data element. When satisfied that all data is correct, hitting dsubmitt enters the validated data to the research database. This validation approach was used successfully to minimize entering ineligible subjects into a large clinical trial [3]. Both approaches show the ability to validate electronic source data for research purposes without requiring paper as an intermediary step. This approach acknowledges the medical capabilities of the investigator doing the research data collection, and the investigator is treated more professionally than in the current system. Certainly, as electronic EMR systems proliferate in the future, paper medical records will become history, and the requirement that data from a paper medical record as the source document will be moot. Ideally, we will have our electronic research systems integrated with electronic EMR systems for efficient and accurate data transfer between the two systems.

9 R.G. Marks / Controlled Clinical Trials 25 (2004) Fig. 6. An alternative validation approach for electronic source data is to enter all data from multiple screens, and then validate the data at one time from a complete text listing of all data.

10 446 R.G. Marks / Controlled Clinical Trials 25 (2004) Discussion The clinical trials industry acknowledges the need to improve the efficiency of their operations. Current time to get a product to market leaves little patent protection time to recoup development costs and make a profit on most drugs. Discussions abound on ways to reduce the drug development cycle to increase market time on patent. Many clinical trial professionals believe it is essential and possible to reduce the conduct time and cost of Phase I, II and III clinical trials through better deployment of electronic processes to increase efficiency and safety of clinical trial conduct. Estimates have been made that 1 2 years can be eliminated from the clinical trials development cycle once clinical trial processes are reengineered, taking advantage of available electronic technologies. In this suggested re-engineered process for capturing clinical data electronically, the investigator s workflow is changed considerably. The investigator will take more time to initially capture the data, as direct data verification takes place as data is entered. The advantage is that monitoring effort is tremendously reduced for the investigator, with the hope that the gain in time savings from monitoring more than offsets the additional time required for initial data capture. In addition to enhanced efficiency from moving to electronic source documents in EDC, other clinical trial processes are ripe for enhancing from electronic capabilities. Examples are IRB work, investigator training, randomization, study drug management, investigator compensation, and adverse event management. Each of these areas can be integrated into a more comprehensive network of trial workflows, which can provide further efficiencies for study investigators and other clinical trial constituents [3 6]. These efficiency gains are only possible as attitudes change regarding moving from old inefficient manual processes to more automated processes which provide complete documentation of work being conducted under FDA 21 CFR Part 11 guidelines for electronic systems for clinical research. Statement of Financial Interest: Ronald Marks is an inventor of this technology and is involved in its commercialization. Dr. Marks may benefit from this technology by receiving royalties and equity growth. References [1] FDA. 21 Code of federal regulations part 11. Electronic records; electronic signatures; final rule. Federal register, vol. 62, No. 54. Washington DC: U.S. Government Printing Office; 1997 (March 20). p compliance_ref/part11/. [2] FDA. Pharmaceutical CGMPs. Guidance for industry part 11, electronic records; electronic signatures scope and application. Federal register, vol. 62. Washington DC: US Government Printing Office; 2003 (August). p [3] Marks RG, Bristol H, Conlon M, Pepine CJ. Enhancing clinical trials on the Internet: lessons from INVEST. Clin Cardiol 2001;24(Supp. V):V [4] Marks RG, Conlon M, Ruberg SJ. Paradigm shifts in clinical trials enabled by information technology. Stat Med 2001;20: [5] Marks RG, Conlon M. eclinical trials: the future is now. J Commer Biotechnol 2002;9: [6] Marks RG. Potential benefits of web-based technologies in the conduct of caries clinical trials. J Dent Res 2004;83(Spec. Issue C):C25 8.