During the last 20 years, the complexity of clinical development has grown tremendously. Not only has the average number of trials per New Drug Application (NDA) increased from 30 to more than 70, the number of patients in a typical submission has also increased from 1,576 to about 4,200 (from 1980 to 2000).
In conducting clinical trials, pharmaceutical companies traditionally use special three-part NCR forms, known as case report forms (CRFs), to enable investigative sites to record protocol-specified data about patients during the trial. For every large clinical trial, tens of thousands of CRFs must be collected from hundreds of clinical trial sites. The sponsor of the clinical trial typically has no access to the data until it has been monitored by a Clinical Research Associate (CRA), sent in to the central data facility and entered into the database. This typically creates a delay of 8-12 weeks.
Once entered into the database, the data must be reviewed, data correction requests must be issued to the site and their responses reentered and rechecked. Managing this batch process is cumbersome, time consuming and error prone. It requires extensive human intervention with in-house data entry personnel interpreting investigator’s handwriting, double data entry, hand review of data, and multiple cycles of review, and back and forth mailing of data.
Costs to manage CRFs in a typical late-phase clinical study range from $150,000 to several million dollars. The traditional paper-based process adds months, even years, to the clinical development process for an individual drug. Because of this cumbersome process, patients are denied access to potentially life-saving therapies, and each day lost in bringing a drug to market can cause a significant loss of revenue to the drug manufacturer.
In conventional data management applications, data is locked away in data repositories, for real-time intervention, data review, or clustering. Any analysis of the data requires transfer to programs such as SAS, where data can be properly reviewed. The process of data cleaning involves laborious manual and semi-automated edit checks and data reviews, leading to requests for data corrections from the individual sites. By its very nature, the paper-based clinical trial data collection process fails to collect important metadata (that is, data about the data) that sponsors can use to analyze the efficiency of the trial. Examples of metadata include: Who entered the data? When was it entered? Who modified it? Why?
Most people who have experienced the process of data collection and management in clinical trials view it as an aging and difficult process that is in need of significant updating.
Enter Information Technology
The inefficiencies in clinical trial data management have led to many technology interventions during the preceding 20 years. Until recently, the available technology was not an adequate match for the process needs. Earlier attempts at automating the process, such as fax and traditional remote data entry (RDE), failed to gain momentum and remained in the pilot phase at most companies. However, the recent arrival of Web-based clinical trials is creating a quiet revolution in the clinical trials industry, with a rapidly growing percentage of clinical trials being performed on the internet. A brief look at the limitations of fax and RDE is valuable before engaging in a discussion of Web-based clinical trials.
Fax and RDE
The fax machine combined the electronic transmission of data with the paper process. This attempt to speed data collection eliminated the delay between initial data entry and CRA monitoring. Theoretically, data could be reviewed almost immediately after a patient was seen. However, faxed data requires either optical character recognition, which is frequently inaccurate, or manual data entry. Moreover, the fax does not allow the capture of any of the important metadata. Fax-based systems only answer a small part of the inefficiencies of the clinical trial data management process.
Remote data entry was the next attempt at speeding the trial process. In theory, RDE also could eliminate the delay in access to data, as well as the need for data entry. Furthermore, it was possible to collect some metadata through RDE software. However, RDE requires a computer, a database and software to be located at the site, which, in turn, creates a significant administrative burden for the site. Sites need to lock up their computers, create frequent backups, have a single computer for each clinical trial and other system tasks. In addition, the data uploading process through dialup lines frequently failed, allowing data to age for weeks to months before it was uploaded. Until data was uploaded, it was susceptible to being lost through damage to the file, theft of the computer, or other technology failures. Many of the potential advantages of RDE never materialized because of these issues.
Enter the internet
The next step in electronic data capture was the Web-enabled clinical trial. The Web solved the most difficult problem for electronic clinical trials—how to create a robust data network at tens to hundreds of clinical sites in a matter of weeks. A data network could be set up between the time the site was identified and the start of the clinical trial. This type of data network fit the purpose of the internet perfectly: allow diverse, geographically dispersed sites to communicate through a single network, regardless of the underlying local infrastructure.
Web clinical trial applications can be developed with two underlying philosophies: the first is a pure internet browser-based application that operates in "online" mode with no local specialized software or data. The second is a modification of traditional RDE—the specialized local application that operates through the internet. While bandwidth and internet availability issues may limit "online" modes in some locales, the advantages of the internet can be lost in some of the offline and "hybrid" Web clinical trial applications.
Web-Enabled Trials: More of the Same
Since any software that communicates using the TCP/IP protocol is capable of working over the internet, it is fairly straightforward to convert an existing software application into a "Web-enabled" program. This approach modifies an existing application to send data through the internet by adding on a Web component. For example, a program might work entirely offline and synchronize regularly to a server via the internet. Alternatively, a stand-alone program can be retrofitted to operate within a Web browser, giving the appearance of being a Web application, but operating with local software and data.
All of these solutions will allow an application to transmit data over the internet. However, the application remains a remotely installed, stand-alone client/ server application. It will require software on the remote computer, updates, maintenance, validation, restricted computer usage, and will have all of the risks of corrupted or stolen local data found in traditional RDE systems. Data transmission delays eliminate many of the advantages of online clinical trials. In addition, many of these offline or hybrid programs use Java or ActiveX technologies, which create significant issues for validation, security, and often are excluded from corporate and site firewalls.
Web-Based Trials: An Optimal Solution
With online internet technology, pharmaceutical companies can capture clinical trial data in real-time. Web-based clinical trials bypass the queuing and waiting that is inherent with paper-based trials, fax, RDE and "Web-enabled" solutions. Data from electronic CRFs are sent over the internet to a central database as soon as they are entered. During the data entry process, queries are handled immediately, so that data are cleaner, earlier.
A true Web-based product is designed specifically to work with local Web browsers, with no local data or additional software. It has an underlying Web architecture that optimizes real-time internet clinical data exchange. In addition, real-time alerts are enabled, allowing intervention in the case of serious adverse events, or automated requests for drug shipment. Since the data are all centralized and there is no data stored locally at the Web-based site, the administrative burden is almost eliminated at the site. In addition, there are no data security issues at the site since there is no local data.
Web-based products must also guarantee data integrity through the use of transaction processing (TP) monitors. TP monitors coordinate the movement of data through complex operations. They allow multiple parallel database entries while simultaneously guaranteeing the completeness of the data. For example, when one submits a case report form to the database in a Web-based system, automated edit checks and queries could be triggered. TP monitors guarantee that both the edit checks and the data are written to the database. Neither could be entered alone, thus preventing logical database corruption. TP monitors have been used in financial service applications for more than 20 years to guarantee that debits and credits always match.
For the next few years, there are some valid reasons why sponsors might want to distribute clinical trial data and applications for "offline" use. They may want to collect clinical trial data from distant locations, such as rural Africa, where there are weak communication infrastructures, or where communications between continents is not optimal. Sites may also want to use electronic data entry for the capture of source data, which requires a local copy of the data at site. In addition, CRAs or sites may need to work in operating rooms or other locations where internet connectivity is not available.
Web-enabled solutions for these problems involve offline or hybrid architectures, which require putting an application and data collection program out at the individual sites. The data collection program works over the internet so that people can work either online or offline. These systems are usually built on Microsoft Windows 98 operating systems that are severely lacking in security features. In addition, they typically operate with limited database tools, such as Microsoft Access, and also lack TP monitors. The result is that they are fragile, prone to errors and failure, as well as data loss.
The requirement for offline clinical trial data entry is best met with a network of distributed systems that replicates, or updates, clinical trial data at remote locations. The servers each contain a complete, robust clinical trial application, built with the security of the Microsoft Windows NT® or Windows 2000 operating systems and incorporating a complete, enterprise-class database with TP monitors.
Full data exchange occurs both ways between the individual servers. To run clinical trials on a global scale, distributed servers can be strategically placed on appropriate regions to avoid communication delays. Each server can act as a hub for communication with local computers via the internet or private networks. When individual offline data entry is required, an individual laptop computer could act as a distributed server, essentially providing offline capabilities. Distributed systems also enable point of care wireless data entry, and offline data review.
Distributed servers require intelligent, application-based synchronization of data. Synchronization enables the merging of changes from various distributed servers with resolution of data conflicts (e.g. two changes to a single piece of data from two locations; two answers to a query, etc.). Conflict resolution must be as automatic as possible, but requires the software to incorporate heuristic rules from clinical trials to perform this properly.
In an automatically synchronized distributed server system, data may be out of sync by only minutes, providing a negligible delay in data availability and maintaining the real-time trial management features of Web-based clinical data management. In addition, data synchronization through distributed servers can be used to create nearly real-time backups and data archiving through trusted third parties.
There will always be some issues to face as companies add new technology and change their business processes. Currently, some of these issues involve integration with in-house databases, internet performance and connectivity, and security.
In-house Databases: Typical in-house clinical data management systems act as static data repositories, which are more like libraries than databases. Optimally, data are collected from a clinical trial into a transactional database, designed for direct export to analytic tools like SAS software. In parallel, the data and metadata can be ported, using XML data tagging, into a data warehouse and data marts, where it can be mined for valuable insights, combined with other data, and used for operational management of a clinical trial or clinical development program. The usefulness of traditional, static data repositories is waning.
Performance and Connectivity: Perform-ance and connectivity have changed dramatically and continue to change for the better. More and more computers are being made available at clinical sites for data management applications as the cost of hardware continues to go down while performance rises. In the U.S., connectivity to the internet is generally not an issue. In other countries, internet connections are increasingly becoming available and more institutions are using them for communications.
Bandwidth: Bandwidth refers to the transmission capacity of a network medium and is measured in bits per second. The more information to be sent in a given period of time, the more bandwidth required. As bigger bandwidth is created, there are always bigger bandwidth applications waiting in the wings that tend to slow down the internet. But if Web clinical trial applications are designed with small page sizes, the increasing bandwidth capacity of the internet will continue to allow more efficient, effective performance at investigator sites. The key is to design the Web software so that pages quickly download and execute.
Security: Good security means two things: safeguarding data from theft, and guaranteeing the quality of the data. Safeguarding data is an issue for all types of computer systems and databases today, whether they involve in-house manual data entry or internet-based data entry. If the data is accessible on a local area network within a company, it is already typically exposed through corporate internet connections, and dialup modems. When properly designed, a Web-based clinical trial data management solution can be extremely secure. Unlike networked data, the Web-based clinical trial server can be isolated from all other servers, behind a firewall that is constantly monitored. There should be no back door access to the database through modems and the server should be located in a physically secure location. In addition, there is no local data at the site, and when 128-bit secure socket layer (SSL) encryption is used, the security matches that used for transmission of top secret documents by the U.S. government.
The second point of data security involves the actual software application and how it manages the data. Some applications use databases and forms of communication that are fragile and insecure. Frequently, these systems can lead to corrupted databases, forcing sponsors or CROs to retreat to earlier backups, losing hours to days of data that has been entered. The optimal solution is to use validated, robust clinical trial applications that incorporate TP monitoring and are frequently backed up for offsite data storage.
Increased competition and regulatory demands are powerful forces pushing pharmaceutical companies to speed up their clinical trial processes. Technology solutions for data capture and management are an obvious source of efficiencies in clinical development. Web clinical trials offer a significant step forward in the repair of an aging and increasingly inefficient clinical development process.