64S Abstracts

replace current systems. A graphical interface written in the widely used hypertext markup language (HTML) and using common gateway interface (CGI) scripts allows researchers to easily fill in patient information, check on eligibility, and receive allocations through an internet browser such as Netscape or Internet Explorer. There are numerous advantages to computer based assignment, including access from any networked computer, availability around the clock and customizable levels of security. With some careful planning the computer based system can be made as or more secure than the current methods of distributing randomizations. Security issues such as limiting access points, password protection, and use of detailed access logs are discussed. Implementation issues such as compatibility of computer platforms and browser versions, handling multicenter trials and stratified designs, and dealing with simultaneous requests for randomizations are also discussed. Researchers conducting clinical trials at Scott & White Memorial Hospital now have the option of receiving randomizations over the internet. The use of computer based systems for randomizations will spread rapidly due to the ease of use and management.

49 BREAKING INTO DATABASES: INTERACTIVE

DATA ENTRY ON THE WEB

Sharon M. Bak, Suzette M. Levenson, Tom Duria and Jim Hill Boston University School of Public Health

Boston, Massachusetts

As the Data Management Center (DMC) for clinical studies for cocaine pharmacotherapy with the National Institute for Drug Abuse and Department of Veterans Affairs, our role has been to facilitate communication among the five Medication Development Research Units (MDRUs), to obtain agreement on content, format, and procedures necessary for the standardization of data management, and to monitor study progress. In order to efficiently track enrollment and provide progress reports for the many small studies being conducted at each of the MDRUs, we have enhanced our existing web site by adding an interactive database system.

Using Microsoll Access TM and Visual InterDev TM software, we created "active" web pages enabling the MDRUs to enter basic demographics of participants as they are enrolled, and retrieve enrollment summaries across all studies and sites. For security, all data transmissions were encrypted using Secure Socket Layer (SSL) protocol.

Utilizing commercial software and Open DataEase Connectivity (ODBC), our static pages were converted to dynamic ones, resulting in an integrated system for trial management and data entry.

50 BALANCED APPROACH IN CLINICAL TRIAL MANAGEMENT (CTM) BETWEEN MODERN

TECHNOLOGIES AND PRACTICAL REALITIES

Thomas Caldwell, Rex Welsh, Jean Owen, JoAnn Stetz, Brenda Young and Thomas Pajak

American College of Radiology Philadelphia, Pennsylvania

In moving to a "paperless" system the ACR is developing a new methodology for handling and processing data collected via the World Wide Web in the administration of clinical trials. The ACR Data Center will accept the data in the form of paper, FAX or electronic. The ACR website will provide the capabilities for registering patients via completion of electronic forms available on-line. Paper and FAX clinical records will be converted to electronic records. The first such trial to use this technology will receive clinical data and MRI images

Abstracts 6 5 S

in digitized form for a clinical trial examining the efficacy of MRI in diagnosis of Breast Cancer. On-line computer editing will identify missing information and perform range cheeks for inconsistencies. Data manager review will be performed on the electronic data submitted. Diagnostic MRI images will be electronically distributed for review by "expert" radiologists on their local PC's with results returned via electronic forms. The fmal data will be sent to the Statistical Office for analysis. Besides managing individual data forms, the study also will determine the mechanisms essential in accomplishing efficient transfer of information between the investigative institutions, Data Center, Statistical Office, Chairman's Office and the National Cancer Institute. This approach may aid in increasing patient enrollment in cooperative group trials by providing flexibility in data management requirements. RTOG trials will further test this process.

51 A PASSIVE INTRANET FOR VIEWING IMAGES, DATA,

AND DOCUMENTS AT A MULTICENTER TRIAL COORDINATING CENTER

Stephen C. Grubb The Johns Hopkins University

Baltimore, Maryland

An intranet is an inlbrmation system that is accessed like the World Wide Web (by using a web browser) but is appropriate for internal delivery of sensitive information because it is not accessible from the Internet or the outside world.

An intranet has been developed for use at the coordinating center of a muiticenter clinical trial. Because this trial collects data on forms that are directly faxed from clinical centers to a computer for electronic processing and storage, paper forms are not stored. This system serves as the primary means for data forms to be viewed at the coordinating center, and also provides access to data displays and documents, and facilitates image workflow (the processing and organizing of images). Staff members use standard web browser software, in conjunction with an auxiliary "helper" program that displays fax images, captures user input, and accesses the study database.

The technical development and administration of such an intranet may be simplified if all content is "passive", that is, made up of files located on the same computer system as the users. In situations where this may be done, a passive scheme eliminates the need for a web server program, and reduces concerns related to security and data access control.

This presentation includes an outline of the current content of the Submacular Surgery Trials (SST) Coordinating Center intranet, and screen snapshots of the software application. Also discussed are the intranet's use with regard to forms image workflow and data edit processing, and its use as an archive for documents such as the study protocol manual, manuscripts, clinic notices, and internal documentation.

52 MODELING AND SIMULATION IN THE DRUG

APPROVAL PROCESS

Donald B. Rubln Harvard University

Cambridge, Massachusetts

The focus on frequency operating characteristics of procedures is the hallmark of traditional statistics, and standard methods in the drug approval process are typically designed to be very simple (e.g., a t statistic) so that their operating characteristics can be analytically derived. These standard methods, however, can be inefficient, and more importantly, cannot adequately address complications such as noncompliance or drop out. When faced with such