If asked how computers, clinical trial software, and related technologies are changing clinical trials, most would probably think about the efficiencies that these technologies bring to the process. Faster data entry, more efficient communications between participants, and improved process management would be what springs to mind. While these are real and important benefits of clinical trial technology, they miss a key, driving benefit-changing the design of the clinical trial itself. (I am using the term "design" to cover both data- and operations-related factors.)

As I have discussed in previous columns, technology is the great enabler of process change. The value that any technology brings is often related to the creativity that users bring to the process. Through a thorough understanding of the capabilities of a given technology or set of technologies, creative people can change the fundamental process that the technology is applied to. A real-world example may help demonstrate this.

Not very long ago, our options for watching a movie or television program were limited. We could see the films that played in the movie theater at fixed times or watch television programs when they were broadcast.

New technology has led to a process change in movie distribution. The emergence of videotape, followed by DVD, made it possible to buy or rent almost any movie you wanted by going to the store. The arrival of the Internet enabled another leap forward--NetFlix created a business where anybody could order movies from an online database to be shipped for watching at home. Most recently, a Web site called Cin-o-matic¹ has made it possible to look at reviews from hundreds of movies available on video (or in the theater). Any movie of interest can be added to your list of NetFlix "orders" with one click.

Beyond this, the availability of broadband in many homes makes it possible to have movies distributed to your home immediately, without leaving your home, by your cable company.

In the course of 20 years, technology has enabled a revolutionary change in the process of watching movies. Yet, none of the advances in technology were created specifically for the process of viewing films--they were created for general purposes such as with the Internet, or for related purposes such as with cable. It was the creativity of people, using the enabling technology, who created these remarkable ways that we can get movies. Now, let's turn to clinical trials and technology.

Clinical trial design and technology
When considering clinical trial design, most of us think first about clinical endpoints used for statistical analysis. There are several important ways in which technology enables a dramatic change in clinical trial endpoints.

First is the collection of clinical endpoint data directly from the patient, today known as electronic patient reported outcomes (ePROs). Most clinical endpoints are collected directly by a physician or study nurse--accurate, objective reporters of quantitative data who understand the principles of scientific study. Certainly, for the measurement of blood pressure or ECG changes this makes sense.

However, the clinically meaningful endpoints for many clinical trials are observations that the patient makes and reports to the physician. For example, number of cigarettes smoked, frequency of headaches, episodes of nausea, etc. We have come to accept that these endpoints are collected at periodic, scheduled clinic visits during questioning by study staff. Patients are notoriously inaccurate at collecting data into paper diaries, and the data are usually of insufficient quality to be used as a clinical endpoint.

2 has made it possible to consider collecting clinically meaningful endpoints directly from the patient, through the answering of questions presented to the patient on a handheld diary device. Compliance with these devices and the quality of the data may be sufficient for it to be used as an approval endpoint for certain indications.

Another form of data collection from patients involves the use of biometric devices. These can be simple, such as an electronic spirometer that records the time, date, and measurements of home-administered pulmonary function tests. There are now devices available that can simultaneously record dozens, if not hundreds, of physiological variables, such as the LifeShirt from Vivometrics.³

The collection of physiologic data creates certain challenges for clinical trials. For example, if physiological measurements are used as endpoints, it may be necessary to consider different statistical measures than are commonly used for evaluating the efficacy of a drug. The data are often "messy" and are continuously variable. While statistics exist for evaluating this type of data, they may be less commonly used for the evaluation of endpoints in clinical trials.

In addition, the availability of dozens of different measurements may provide for apparently significant results that do not bear out in repetition. With gold standard endpoints in use for many diseases and indications, it will be necessary to clearly demonstrate that these surrogates effectively predict the clinically meaningful outcome.

Finally, it is likely that continuously monitored physiological variables may identify anomalies that may or may not represent adverse experiences. How to distinguish asymptomatic physiological variations from adverse experiences may create some unique challenges.

One aspect of clinical trial design that should not change much with the use of electronic CRF technology is the amount and type of data items that are collected. CRF design may differ (eCRFs may not have the look and feel of paper CRFs), and subtle differences may be apparent (such as check boxes indicating choice now replaced with "radio" buttons onscreen), but for the most part good statistical/data management design of a CRF page should be generally applicable to the eCRF.

The temptation to load up CRFs with data items because of the reduced cost/distribution hassles of the eCRF should not lead to over-collection of data. Data collection still needs to be limited to the data items required for analysis of the (hopefully) targeted efficacy and safety measures of the trial. In addition, the ability to provide a large pulldown list of options for a question where such a list wouldn't be available on paper may be helpful for certain purposes (e.g., coding). However, the availability of a even a large list introduces the possibility of bias.

There is little question that eCRFs have the ability to improve the quality of the collected data. For example, typographical errors and out of range data can be caught immediately, on entry. This type of feedback provides the best way to educate sites and reduce the occurrence of the error. Of course, for any questions that could introduce bias, it is essential to have a full audit trail of the query. In any case, all data changes should have a full audit trail.

Another area where technology can change the overall design of a clinical trial is in the statistics of randomization and power. Using an EDC system or ePRO system, it is possible to design a trial that is automatically modified as it progresses through an adaptive trial design.

One well-known form of adaptive trial design is "play the winner," whereby a predetermined outcome is measured and randomization is biased through a predetermined formula towards the "winning" therapy. The benefits are that fewer patients are exposed to the less effective therapy, presumably more safety information can be collected from the more effective therapy, and fewer patients may need to be studied overall before determining the statistical and clinical significance of a therapy.

When outcome data is collected electronically, it becomes possible to create a loop from the outcome to the randomization, so that the "play the winner" strategy is applied efficiently, without human intervention. Of course, since this requires that randomization information be placed within the data collection system, special care must be taken to avoid unblinding and bias. But proper management of technology and processes can prevent this, and provide appropriate assurance to regulators.

A second type of adaptive trial design involves ongoing assessment of the sample size, to avoid under- or over-allotment of patients. For example, if the statistical power of a trial is based upon a particular variable and an estimate of its variance, it is easy to see how an increase or decrease in the variability of the sample could affect the power. By continuously monitoring such a critical factor, it is possible to adjust the sample size of a trial for the power that is desired. Again, this can be done with a manual, people-intensive process, but is ideal for automated feedback from EDC-collected data. At least one software product is built to directly address these, and other types of adaptive trial design.⁴

Clinical Trial Logistics and Technology
In designing a clinical development plan and a clinical study, it is often essential to keep in mind the availability of study drug. This is especially the case for a biotechnology company, where the production plant may be highly specialized and expensive to build. A small pilot plant may be all that is available for supplying an entire study. Furthermore, stability data may be incomplete in early Phase II. Many companies limit enrollment or the number of centers involved in early studies because they are unable to supply an adequate amount of blinded study drug to a large number of sites.

This is one area where technology can enable a significant change to study logistics. Patient enrollment data entered into an EDC system can trigger automatic, electronic (email or fax) notification of a study drug fulfillment center, along with pertinent information about the patient. The proper study drug, perhaps adjusted by patient BMI or weight, can be shipped out overnight to the site that enrolled the patient--on a patient-by-patient basis. Again, such a system is possible without technology, but is labor intensive and subject to failure. The enrollment of patients within an EDC system enables logistic changes to the conduct of the study.

Perhaps more mundane, but nevertheless important, is the ability for EDC to enable size and location design decisions that could otherwise be impractical with paper. In large, simple trials and very large Phase IV trials, the logistics and costs of shipping paper and the costs of monitoring travel become prohibitive for studies. Often these factors are limiting in the size and scope of such studies.

With computers and Internet access essentially ubiquitous in medical settings, the logistic simplicity of eCRFs and the ability to monitor data in near-real time at a distance enables the conduct of studies in large number of centers and in remote locations.

Clinical trial site management
Part of every clinical trial plan and design are the decisions made on how to identify sites, communicate with and train them, monitor their progress, and secure their best performance. Design decisions about the trial include number and location of sites; both can be influenced by cost and logistical considerations.

Previously, these have been mostly person-to-person activities, requiring travel by the site or the monitor, cost, and time away from office. A number of excellent technologies have been deployed to help in the site/sponsor interaction, and together these can help achieve higher quality data in a shorter period of time. Many, if not most, studies today have document portals where start-up documents can be downloaded, completed 1572s and other site documents can be shared with sponsor, newsletters communicated, and the entire study folder process managed.

In addition, these portals are often used for live investigator training and training/testing of site personnel in study procedures. Often, portals are being used instead of the traditional investigator meeting, saving costs and time.

One of the most powerful incentives for better performance is to allow people to see their performance compared with others. By taking information from an electronic data capture system and displaying it to sites, it is possible to give them an up-to-date view of their activities compared with other, anonymous sites.

For example, sites can be graphically shown their recruitment, time from query to response, queries per page, and many other metrics. These reports will allow the site to understand areas that need improvement, or areas in which they excel but can be set up to avoid disclosing the identity of the data from other sites. Technology, in this case, enables sites to achieve their best performance in collecting high-quality data in a clinical trial.

Conclusion
Technology enables changes in clinical trial design in major and small ways. It is certain that current technologies and emerging technologies will continue to enable the evolution of clinical trial design. The gating factor for these changes is the ability of each of us to understand the value that a technology can bring, and the creativity to apply it to the design and implementation of clinical trials.

References
1. http://www.cin-o-matic.com
2. A.A. Stone, S. Shiffman, J.E. Schwartz, J.E. Broderick, M.R. Hufford, "Patient Non-compliance with Paper Diaries," BMJ, 324, 1193-1194 (2002). Available at http://bmj.bmjjournals.com/cgi/content/full/324/7347/1193
3. http://www.vivometrics.com
4. http://www.cytel.com/East/Default.asp