The Need for Electronic Data Capture in Phase I

Pharmaceutical companies and clinical research organizations have been working to improve productivity and return on investment by modifying processes, decreasing cycle times, and implementing improvements wherever possible.¹ According to the FDA, it is during the clinical trial process that 9 out of 10 experimental drugs fail.² Recent research by Joseph DiMasi of the Tufts Center for the Study of Drug Development suggests that reducing the length of clinical trials by 25% would save $129 million per drug. ³

Phase I studies are the starting point for drug trials in humans and are a vital stage of clinical development, where critical decisions are made with regard to whether an investigational drug proceeds into Phase II efficacy trials or is halted. Rapid, scientifically based decisions not to proceed can achieve cost savings and redirect financial resources to more promising drug candidates. Numerous technological innovations have recently begun to automate Phase I trials, but one critical component remains virtually untouched: effectively integrating clinical lab data directly into studies to improve both safety and speed.

Currently, most Phase I clinical research centers receive clinical lab results on paper, which requires intensive and time-consuming review by staff and opens test interpretation to human error. Many of these test results are used to assure subject safety, so timely review is extremely important. In addition, sponsors often request electronic data transfer of clinical results at the end of the study. Preparing electronic data from paper reports is frequently a cumbersome, time-consuming, error-prone process.

This article discusses why standard clinical lab systems are generally not well suited to the process of Phase I clinical trials, and why an electronic system that is tightly integrated with the operations of a Phase I clinic is essential for process improvement and, most importantly, timely evaluation of results.

Phase I clinical lab needs

Test results from clinical labs are used extensively in Phase I clinical trials, primarily for screening and safety assessments, and increasingly as biomarkers. For Phase I clinics with more than 100 beds, it is often cost effective to establish an on-site clinical laboratory. These labs are likely to deploy software systems that automate laboratory processes and store test results in a database.

Most clinical lab software tends to focus on the needs of medical facilities whose emphasis is on rapid testing for diagnostic evaluation and billing automation. These systems do not readily facilitate the timely review of test results from a subject safety or inclusion/exclusion perspective, nor do they provide capabilities for database creation for statistical analysis purposes—essential features for Phase I research. Significant advantages could be realized with a lab information system that tightly integrates with the processes of Phase I studies (see Figure 1).

Figure 1. An EDC system better suited for Phase I studies would capture inclusion/exclusion rules, potentially saving the PI time. In addition, automatic alerts would increase the safety of subjects.

Features are needed that would allow test definition and sample processing based on individual study protocols. Integration of test results with automated alerting could reduce the human resources needed to process the lab data. It could also facilitate a more efficient review of the clinical lab results in the areas of qualifying subjects, evaluating safety results, preparing adverse event (AE) documentation, investigator result sign-off, and data audits. Improved study safety for subjects would be achievable with real-time alerting based on test results. This would allow for timely consideration of whether to continue dosing, escalate dosing, repeat dosing, or stop.

EDC for Subject Eligibility

Differences in clinical lab system needs

Typical clinical lab software is focused on the needs of medical clinics (hospitals and physician offices), which are very different from those of Phase I clinics. For a medical clinic it is of great importance to assure that each test meets Medicaid-defined billing rules based on the patient's diagnosis. Since most Phase I facilities do not bill clinical tests based on patient diagnoses, these rules are usually unnecessary and tend to confuse the standard operations of the Phase I clinical laboratory.

Medical clinical labs must always test a sample based on all ordered test panels. For Phase I clinical labs, a cost savings could be realized if testing for a subject was canceled based on the results of one test. For example, a confirmed positive pregnancy value or positive urine drug screen could trigger a study exclusion rule. In this case no additional testing would be required, as the subject would be excluded from the study.

In most medical centers, lab test methods and patient alert limits are defined once for the whole facility and are rarely changed. In a Phase I clinic, lab tests, panels, test methods, and patient alert limits may change from protocol to protocol.

Integrating information & study processes

Phase I protocols often require sample collections at the initial screening and at various time points during the initial and subsequent dosing. An integrated clinical lab system could automatically order these samples based on the study protocol so that collection and processing would be more efficient. Preprinting of barcoded collection labels would also speed collection and processing. Barcode scanning could automatically check the IDs of subjects and collection, sample, and transfer tubes—virtually preventing human error.

Once the subjects are admitted for a Phase I study (usually 12–24 hours prior to dosing), additional samples may be collected for testing. This ensures that inclusion and exclusion criteria are met for each subject and catches recent changes that might affect the subject's study eligibility. In addition, the results may serve as baseline values against which to evaluate any postdose laboratory result variations. The Principal Investigator (PI) makes a final decision to dose based at least in part on the review of a paper-generated report. In a tightly integrated electronic system, automated alerts could notify the PI of subjects that fall outside the study limits. An electronic report with appropriate search and sort capabilities would improve the speed of subject review and approval. Just prior to dosing, an additional automated review would have the potential to improve subject safety by capturing any potentially late arriving test results.

It is common in Phase I study protocols to schedule sample collections after the initial dose for additional clinical testing. This testing assures the safety of the subjects and satisfies the criteria for continued dosing. If test results are available electronically, automatic alerts may be generated to reveal subjects who are exceeding the clinically defined safety ranges. With automated alerts, subjects whose test results exceed safety limits can be quickly evaluated for a potential AE and treated. Ranges of test values, along with the medical judgment of the PI, determine if the AE is rated as mild, moderate or severe. An electronic system could produce reports to show all lab values related to a particular subject experiencing the AE. This could both improve subject safety and the reporting of AEs by alerting clinicians in a timelier manner.

Further alerts could identify subjects no longer qualified for continued dosing based on protocol-defined test limits. These alerts and specially formatted reports would decrease the time needed for subject evaluation. Finally, just prior to dosing, the subject's test results could be automatically checked and flagged if out of range. This would improve safety and assure compliance with the protocol's dosing rules.

PIs need timely and thorough documentation and assessment for each AE that occurs throughout a clinical trial. This documentation and assessment is critical in helping to identify clinically significant trends that may affect subject safety. The PI determines if the event is related to the test compound, and typically a classification of the event is recorded. Current systems usually involve a hard copy of the AE form passing through several hands, which may delay the assessment process. A tightly integrated lab information system would send electronic AE test results directly to research staff and sponsor, and would include the results in a final AE report.

In an escalating dose study, "stopping rules" may be associated with one or more clinical laboratory test values exceeding a range defined by the protocol. These values are based on samples taken from the subjects after receiving the dose. A tightly integrated electronic system could generate automatic alerts for these values so that the PI and research staff could improve the speed with which safety assessments are made. Likewise, appropriate electronic forms with search and sorting capabilities could speed the process of evaluation when a dose escalation has been completed. In addition, the ability to generate subject-specific laboratory result and/or AE reports would aid in the evaluation of trends.

Integrating results from outside labs

Many Phase I clinics do not have internal clinical laboratories. If they do, they often have pre-existing software systems that are incapable of tight integration with the clinic's other electronic systems. In order to integrate test results, it is necessary to import them using some electronic data transfer method. Most outside reference laboratories submit clinical lab results in one or more electronic formats. In many cases this is a proprietary delimited flat file format, which is commonly referred to as a CSV (Comma Separated Value) file format. More advanced formats used include HL7 (Health Language 7) and CDISC formats from the Clinical Data Inter-

change Standards Consortium.

All electronic transfer methods must include the coded test names, test results, and the testable and clinical action ranges associated with the test method. During the result transfer a properly functioning automated system would automatically crosscheck the testable and clinical action with the range values that were approved for the study. Alerts should be generated if the range values do not match. Alerts and appropriate audit reporting for incorrect range values have the potential to reduce approval times.

Although importing test results electronically is by far the preferred method of integrating clinical lab results, it is not always possible to do so. Some reference laboratories do not offer test result exports in electronic format, and realistically there will always be some small volume of esoteric tests for which it would not be cost effective to create an electronic interface. For this reason there will always need to be a data entry screen for tests and their associated testable and clinical action ranges. Alerts for ranges not matching those approved for the clinic or study are necessary here as well.

Lastly, electronic files usually do not include investigator comments for out-of-range lab tests, but this information needs to be captured in the clinical database. Having this information entered by the PI or research staff at the Phase I clinic into a tightly integrated Phase I clinical lab system would be more efficient both cost wise and timewise.

Conclusion

Clearly, any effort to increase the speed of drug development must address the unique needs of Phase I clinical trial centers. Current clinical lab software systems, developed to meet the needs of hospitals and other medical facilities, fail to adequately address the needs of Phase I clinics and are not the answer. Using a Phase I-specific, study-centric, clinical electronic data system would significantly enhance the review of lab results and study execution; expedite study decision making and completion; increase the speed of determining subject entrance criteria; and improve the safety of study subjects.

References

1. E. Pena, "Making Metrics Matter: The Changing Paradigm of R&D Metrics," PharmaVoice, March 2005, 8–20.

2. FDA News, 12 January 2006, P05–06.

3. D. Henderson, Boston Globe, 13 January 2006.

Sherilyn Adcock, PhD, RPh, is vice president and chief scientific officer with CEDRA Clinical Research, LLC, Austin, TX. Michael Willett, PharmD, is president and chief executive officer with Advanced Biomedical Research, Inc. (ABR), Princeton, NJ. John Rosenblum* is president and chief executive officer with Green Mountain Logic, Montpelier, VT, email: johnrose@gmlogic.com

*To whom all correspondence should be addressed.