This document (http://www.WORDesign.com/systemsdesign) is submitted to James Lowe, UC Berkeley Extension Instructor, in partial fulfillment of the Spring 1998 course on Systems Analysis and Design: An Overview (X422)

Introduction
Proposed System Overview
Alternative I: PharmaForms™
Alternative II: Data Fax
Comparison
Summary
References

BerkelyMedLogo.gif (3953 bytes)


Introduction

The Biomedical field is booming, and Tools (databases and software to analyze a biomedical problem) rank number one (over medical devices, heath care services, diagnostics, and therapeutics) for biomedical venture capital funding. We--Daniel Leprince, Preeti Tikekar, Fan Shen, Vijay Patel, Krishna Shroff, and Thomas Albert--therefore establish Berkeley Medical Systems™ to satisfy with state-of-the-art software (PharmaForms™ for "digital reliability in clinical trials"™), hardware, and networking technologies the critical demand for accelerated time-to-market of new pharmaceutical therapies that address such terminal ailments as malignant cancer tumors.

The Food and Drug Administration (FDA), a regulatory agency within the United States Department of Heath and Human Services, oversees the process by which a pharmaceutical company (drug therapy sponsor) brings a new medication to market. The journey from laboratory discovery to routine prescription can take over 15 years and cost up to three hundred million dollars. "Each one day delay in a 12-year testing cycle can cost the sponsor more than $1 million."

Our solution, PharmaForms™, address the Clinical Research phase.

phasesofdrugdev2.gif (21842 bytes)

PHASES OF DRUG DEVELOPMENT

  1. Laboratory discovery/development of a promising compound
  2. Pre-clinical research: up to 6 years evaluation using cell cultures and animals
  3. FDA approval to commence clinical research
  4. Clinical research: up to 8 years evaluation
  • Phase I study tests the safety of the compound at recommended doses
  • Phase II tests efficacy of the drug as a cure or therapy
  • Phase III study measures short- and long-term side effects

Afterwards,

  1. FDA review of New Drug Application (NDA):approximately 1.5 years
  2. During review, the FDA validates the study
  3. At its discretion, the FDA can audit the study and make on-site visits to the sponsor
  4. FDA approval

Berkeley Medical Systems PharmaForms are a proposed solution to the high cost, frequent errors, and unnecessary delays in the clinical process. Accelerating the time to market of critical drug can mean dozens of saved or improved lives, as well as revenue to the sponsor of up to one million dollars per day. This proposal follows FDA Guidelines for electronic forms in clinical research.

Diagram0d.gif (13956 bytes)

Explanation of the ten steps in Diagram0

1. Add new patient
The sponsor (the pharmaceutical company) relies on doctors or a contract research organization (CRO) to recruit patients willing to undergo the study. Recruiting subjects for clinical trials is itself a costly and time-consuming tasks. Terminally ill patients who do not foresee recovery with traditional therapies sometimes agree to participate as subjects in a clinical trial with hopes of gaining better health, a longer life, or at least contributing to medical progress. For the sponsor, more participants strengthens the evidence and the likelihood that the  Food and Drug Administration (FDA) accepts the study's findings. Typical studies can have from 100 to 9000 patients. Studies often a placebo (an inert pill) to a control group in a "blind label" experiment. Patient who received placebo later have the option of taking the proposed drug in an "open label" experiment.

An example of a contract research organization is the Sagar Group (http://www.sagargroup.com).

2. Record patient disease history
Typically, the doctor takes the history, but the doctor can authenticate a history that originates with a nurse or Clinical Research Assistant.

3. Record and update concurrent therapy
Often, patients are taking drugs other than the one under study. Recording concurrent therapies allows the study to

  • monitor the efficacy of the new drug in relation with other medications
  • control for the safety of the patient under these new drug interactions

4. Record tumor information
This is a prime area for human error, especially in busy, if not hectic, hospital settings. Our proposed PharmaForm solution automates the calculation of tumor size and body size
(and transfers tumor data to other forms, if necessary). Additionally, storing the information as digital numbers, rather than hasty handwriting, ensures that recorded information is not mis-transcribed. For auditability, the original paperwork is printed at the clinical site and stored on a server in electronic form.

5. Clinical Research Associate (CRA) validates data
The CRA validates data by going through patients' files and making certain that investigators are following regulations and research
protocols. Sometimes the CRA encounters missing, erroneous, or "out-of-range" data. In these cases, the CRA has to spent time and money to obtain correct information. Each day delay in the testing cycle can cost a top-selling drug company one million dollars in lost revenue ("Oracle Clinical Helps Get New Drugs to Market", Oracle Magazine, April 1998, p. 76). Three days can be lost if CRA must fly to the original hospital--which could be in any country in the world-- physically locate the nurse or doctor there who entered erroneous data (or failed to enter required data), create a paper trail of study modifications, and send via overnight FedEx the correction to study headquarters.

6. Quality Control (QA) tracks CRF (Clinical Report Form)
The Quality Control specialist

  • tracks and summarizes patient visits and treatments
  • acts as the liaison between the sponsor site and the CRAs in the field

7. Data Management validates data and performs quality control screening
The data management department

  • Does double data entry
  • Runs validation program to ensure that entries match
  • Queries clinical sites for missing values (including signature and date), errors, corrections and maintains the record of all such queries
    (Note: This process typically takes up to one hour per error, and entails
    • error recognition
    • verification with the original CRF
    • copies of different colors
    • writing a note to Quality Control explaining what needs to be corrected
    • sending the information to the Quality Controller
    • corrects entries and revalidates them
  • Provides clean data for the BioStat department for analysis and reporting
  • Support the Regulatory department in case of serious Adverse Medical Events (AMEs), which must be reported to the FDA within 48 hours

8. BioStat (biological statistics) analyzes data
Programmers and statisticians analyze data according to study protocols and report using analysis tables and data
listings. Statistical analyses enable the sponsor to state findings about the safety and efficacy of the new drug with a specific level of  confidence. The strength of the statistical evidence often determines FDA approval.

9. Regulatory and Clinical report to FDA

  • BioStatistical department submits to the FDA a regulatory package reporting its analyses
  • Clinical department frames the study findings in a clinical and therapeutic perspective in terms
    of medical actions and potential treatment.

10. FDA approves, responds with issues, or audits the study

FDA evaluation board

  • reviews the body of evidence supporting the New Drug Application
  • can request additional information
  • can send evaluators to clinical sites to review the original Case Report Forms
  • can audit
    • If the FDA decides to audit, every piece of data in the study, including any and all changes or "corrections" to original data, must be retrievable. For example, the sponsor needs to be able to show that on 8/3/98, Dr. Knowit changed the number she originally recorded as as 155 ml to 15.5 ml. (Auditability enables to FDA to ensure that only doctors, and not the sponsor, alter study data.)

     

Data Entry Process

Dual data entry is a process to insure homogeneity between the data recorded on the form and the one transferred in the database. The underlying principle is that two different persons are unlikely to make the same mistake in data entry, and that reconciliation of differences is the best way to ensure integrity and accuracy between the data on Case Report Forms (CRFs) and the data stored in the database. It is widely used in the pharmaceutical industry with Case Report Forms. Two clerks are entering the same CRF information in a data entry library through two data entry screens. A validation program compare data entry 1 to data entry 2. If the observations are identical this information is loaded in the master library, otherwise, the records in the data entry library are edited and the validation program is run again. This process is repeated until all differences are accounted for. One of the benefits of using electronic forms is the elimination of dual data entry because data integrity and completeness is enforced automatically.

dualdata.gif (16354 bytes)

Problems with Current System

  • Common errors include incorrect or missing values (including those that need to be transferred "across" one form onto another), calculations, dates, and signatures
    (especially given arcane formulae, such as
    Tumor volume in cubic cm = (L x W x H) x 0.5)
  • Double-data entry incurs expensive overhead
  • Time is lost locating, researching, resolving, and communications solutions to errors
  • Patients and sponsors suffer from delayed time-to-market

The FDA, aware of these issues, provides guidelines for online forms.

Summary of FDA Guidelines for Electronic Forms for Clinical Research

  1. The sponsor provides the FDA a traditionally signed written document certifying that all individuals using electronic signatures (username/password) intend their electronic signature to be legally binding.
  2. The Electronic Case Report form (e-CRF) records complete information on each trial subject.
  3. An audit trail reveals who made any changes, when, how, and why--with the original data still available--and any modified data flagged with a different color or font.
  4. Personnel who create, modify, or delete electronic records are not able to modify audit trails.
  5. The software automatically alerts the user when data are out of the acceptable range.
  6. The software allows users to add annotations, such as ad hoc information useful in case of an adverse reaction.
  7. The system allows the FDA to audit all study information in both human readable and electronic form.

Information Sources

Berkeley Medical Systems gathered and analyzed information from a wide variety of sources:


Proposed System Overview

context4.gif (10813 bytes)

Usability

PharmaForms are designed to integrate well with the physician’s daily routine, and ease the tedium of reporting on patients and treatments. Most cancer specialists have an isolated cubicle where they do paperwork and make phone calls. These cubicles are ideal for establishing internet connections.

How Departments Work Together

deptinfoflow.gif (19188 bytes)

Entity Relationship Diagram

Berkeley Medical Systems™ designed the PharmaForm™ database with a data dictionary customized with tables and many-to-one relations that manage clinical data collection and verification.

er2.gif (10609 bytes)

Data Dictionary Tables

Data Type: N = Numeric, A = Alphanumeric, D = Date MM/DD/YY

Data Item Data Item Description Data Type Data Length
       
INVID Investigator Identification A 4
INVLNAME Investigator Last Name A 20
INVMNAME Investigator Middle Name A 20
INVFNAME Investigator First Name A 20
INVSTADD Investigator Street Address A 30
INVCITY Investigator City A 20
INVSTATE Investigator State A 2
INVZCODE Investigator Zip Code N 9
INVPHONE Investigator Phone Number N 9
       
PATID Patient Identification A 4
PATLNAME Patient Last Name A 20
PATMNAME Patient Middle Name A 20
PATFNAME Patient First Name A 20
PATSTADD Patient Street Address A 30
PATCITY Patient City A 20
PATSTATE Patient State A 2
PATZCODE Patient Zip Code N 9
PATPHONE Patient Phone Number N 9
PATBIRTH Patient Birthday D 8
PATSEX Patient Sex N 8
PATRACE Patient Ethnicity N 8
PATWEIGH Patient Weight N 8
PATHEIGH Patient Height N 8
       
TUMORID Tumor Identification A 5
TUMORLEN Tumor Length N 8
TUMORWEI Tumor Width N 8
TUMORHEI Tumor Height N 8
       
VISITID Visit Identification A 4
VISITDAY Visit Date D 8
PROTOID Protocol Identification A 11
DSASVLSN Dose Assigned Volume Per Tumor N 8
DSDSVLSN Dose Destroyed Volume Per Tumor N 8
DOSERSNC Dose Discrepancy Reason Code N 8
DOSERSND Dose Discrepancy Reason Description A 50
       
PRISUR ID Prior Surgery Procedure Identification A 4
PRISURDS Prior Surgery Procedure Description A 50
PRISURDA Prior Surgery Date D 8
       
CONMED ID Concomitant Medication ID A 4
CONMEDDE Concomitant Medication Description A 60
CONMEDIN Concomitant Medication Indication A 120
CONMEDDU Concomitant Medication Dose Unit N 8
CONMEDFQ Concomitant Medication Frequency A 20
CONMEDSD Concomitant Medication Start Date D 8
CONMEDED Concomitant Medication End Date D 8

Alternative I: PharmaForms™

Approximately 25 remote sites at various physicians’ clinics enable primary data entry. PharmaForms™ electronic Case Report Forms (E-CRF) support data transfer between clinics and the sponsor via the Web or dedicated ISDN lines. Data, once in the Application and Data Server at the sponsor, are available to various departments for evaluation and analysis via LAN/ Ethernet. We propose two main servers at the sponsor:

Application/Data Server for users:

Image Server:

hardreq4.gif (16115 bytes)

Economic Feasibility

Programming Cost for the Forms

$40/hr for two months

$13,866.00

SAS License

?

?

Sub Total

$130,621.00

GRAND TOTAL

$283,166.00

To reduce costs, we recommend dumb terminals, rather than fully equipped personal computers, at the beginning of the project. We assume that the sponsor’s U.S. and Canadian clinics are or can be connected to the internet. Dumb terminals support the critical needs: identifying the patient and visit; downloading forms for data entry.

The sponsor provides hardware to North American clinics having several patients. The data is stored on a central server that enforces data integrity and security. Standard reports answer Frequently Asked Questions by investigators and send the results to the clinics. Database personnel respond to special requests from investigators.

Encryption and Data Security

After evaluating two competing products, Defensor and Entrust, we chose Entrust, the simplest solution with the least training for clients. The Entrust public-key infrastructure (PKI) software provide adequate yet affordable security through encryption for the sensitive information relating to PharmaForms™. Entrust key management is automatic and transparent to the user. This solution is fully scalable for our future needs.

Information about Entrust users is securely archived in the Entrust/Manager database. Entrust/Manager also acts as the Certification Authority (CA), issuing all X.509 public-key certificates. The main functions of Entrust/Manager include:

All security-related events that occur in Entrust/Manager are recorded in audit trails. These records are time-stamped and contain information such as the identity of the person causing the event. This provides integrity control as the audit log files are protected and cannot be manipulated without detection.

This directory contains the name of each person in the organization and acts as a repository for public-key certificates. Administrator’s are responsible for creating a directory entry for each Entrust user. Entrust uses this entry to store the user’s encryption public-key certificate in the Directory.

Entrust/Admin is the main administrative interface to the Entrust system and is used by Security Officers, Administrators, and Directory dministrators. It can be accessed securely over a network or on the workstation that hosts Entrust/Manager.

Entrust/Client allows users to encrypt and digitally sign files decrypt files and verify digital signatures create signing key pairs receive key updates on a regular basis validate certificates before encrypting files for other users or when verifying digital signatures on files securely delete files.

These functions are accessed through an easy-to-use graphical user interface (GUI) that makes key management transparent. Entrust/Client also provides other valuable features, such as the ability to manage frequently used lists of recipients, and the option to save disk space and manage potected files with file compression and archiving.

PharmaForms™ Clinical Intelligence™

We built Clinical Intelligence into PharmaForms™ to minimize the possibility of human error

To enforce data integrity we employ

We designed electronic case report forms using Visual Basic and incorporated formulas that faciliate data entry and significantly reduce the possibility of human error.

For example, the New Patient Screening form

3form.gif (7861 bytes)

COST-BENEFIT ANALYSIS of Electronic-Case Report Forms (E-CRF)
A sample of 200 queries were evaluated individually for their content from a pool of 939 queries associated with an indication of 65 patients. A total of 332 corrections were requested from the clinics. Often more than one correction is requested per query. By decreasing order of importance, problems were associated with missing or unwanted values (63%), errors in transcribing or comparing same variable value across forms and visits (16%), wrong calculations

(10%), values unreadable (4.5%), values outside acceptable range (2%), and others (4.5%).

E-CRF eliminates 41.2% of the queries problems in the first wave by capturing

and associating these with the appropriate record.

For auditing, each original or modified record would be automatically associated with

Other data entry errors are eliminated:

71% of all queries could be avoided with the use of E-CRF

73perc.gif (6743 bytes)

In a second wave of improvement with E-CRF, visual and audible warnings associated with missing required fields prompt the user attention to populated them. We estimate a 70% reduction of missed fields.

In a third wave of improvements, beyond the scope of this presentation, comparisons of values across forms and visits are dramatically reduced through database querying before uploading the forms.

Calculations of Costs associated with the current procedure
Based on our interview with the manager of Data Management, it takes one hour on average for a clerical employee to file a query that need to be sent to the field (i.e. filing query, copying forms on a different color, writing questions and sending). These queries are generally resolved by Clinical Research Assistant when he or she visits the clinics every 4-6 weeks. Excluding travel expenses, one hour of work is associated with each query. Two packages being sent by express mail for each query bring the cost of each query to $55: $40.00 of labor plus $15.00 of shipping costs. This conservative estimate excludes the costs of the time spent by physicians, study coordinators at the clinical site, quality controller, and double data reentry.)

The reduction of 71% of the queries (n=939) to the field save the sponsor $28,400 for an indication of 65 patients. Since several indications are being pursued simultaneously by the sponsor, we anticipate a similar proportion of savings in other studies.

Hardware for Alternative I

Category

Specifications

Unit Price

Quantity

Total Price

Server

Ultra SPRC II 250 NHz Work Station,

10,000

1

10,000

1 MB Cache with built-in 10/100 Ethernet,

3.1 GB Hard Drive, 16x CDROM,

HP Tape Drive, 15" SVGA Monitor,

SUN Solaris 2.6 Desktop Liocense

Exceed X Server with 5 licenses,

Novell Server with 5-licenses, Ethernet Cards

with Hub

Desktop Machines

Compaq Intel 166 Mhz Pentium Processor

$1,399.00

4

5,596

w/MMX Technology, 32 MB SDRAM

1 GIG Ultra ATA Hard Drive/20x CDROM

15" Viewsonic

56 K US Robotics Modems

Microsoft Windows 95/Office97

Business Addition

LaserJet Printer

Hewlett Packard Laserjet 6Lse Printer

$399.00

2

$800

SOFTWARE

Category

Specifications

Unit Price

Quantity

Total Price

DataFax Licenses

4-user DataFax software license1

5

117,000

Desktop Publishing

Adobe Frame Maker 5.5

$800

1

$800

Grand Total ($)

134,196

NOTE:

1. First Year License Fee. Annual renewal fee is $48,000/year.


Alternative II

datafax2.gif (15377 bytes)

DataFax is a data management system designed to manage paper data forms which are faxed from remote sites to a central computer running the DATAFAX system. DataFax is used by organizations to collect and process patient data from hospitals and clinics participating in clinical research studies. 

DATAFAX Features:

Platforms: Hardware and Operating Systems

Hardware for Alternative II

Category

Specifications

Unit Price

Quantity

Total Price

Server

Ultra SPRC II 250 NHz Work Station,

10,000

1

10,000

1 MB Cache with built-in 10/100 Ethernet,

3.1 GB Hard Drive, 16x CDROM,

HP Tape Drive, 15" SVGA Monitor,

SUN Solaris 2.6 Desktop Liocense

Exceed X Server with 5 licenses,

Novell Server with 5-licenses, Ethernet Cards

with Hub

Desktop Machines

Compaq Intel 166 Mhz Pentium Processor

$1,399.00

4

5,596

w/MMX Technology, 32 MB SDRAM

1 GIG Ultra ATA Hard Drive/20x CDROM

15" Viewsonic

56 K US Robotics Modems

Microsoft Windows 95/Office97

Business Addition

LaserJet Printer

Hewlett Packard Laserjet 6Lse Printer

$399.00

2

$800

SOFTWARE

Category

Specifications

Unit Price

Quantity

Total Price

DataFax Licenses

4-user DataFax software license1

5

117,000

Desktop Publishing

Adobe Frame Maker 5.5

$800

1

$800

Grand Total ($)

134,196

NOTE:

1. First Year License Fee. Annual renewal fee is $48,000/year.


Comparison of Alternatives I and II

Alternative II is a less bold step that would attract clinical management operating in a conservative mode. Alternative I offers larger benefits of time/money economy in the mid- to long-run, but does require greater effort up front. Since very little of the gains expected in Alternative II would carry over to a later migration to Alternative I, and since competitors can opt for a route similar to Alternative I, we recommend the organization commit adequate resources for Alternative I. Moreover, Alternative I inherent scalability, and the guaranteed reduction of paper, serve organizational goals as pharmaceutical sponsors add greater numbers of larger projects.


Summary

We have explained the current process of data management in clinical trials, and seen how two alternative systems address, to varying degrees, shortcomings inherent in paper-intensive traditional (or legacy) operations. We believe the design we show provides a detailed feasibility study, and subject to management approval, should prove itself in initial prototyping and usability tests as warranted for the attainment of organizational goals.


References

$532 million in 1997 in Silicon Valley alone ("Expertise Critical in Daring Field of Biotech Investing", Scott Herhold, San Jose Mercury News, Sunday, May 17, 1998, pp. 1D., 4D.
"NTC: Oracle Clinical Helps Get New Drugs to Market", Oracle Magazine, March/April 1998, p. 76., and
Information Technology for Management: Improving Quality and Products, Efraim Turban, Ephraim McLean and James Wetherbe (New York: John Wiley & Sons, 1996), p. 19.
"Guidance for Industry: Computerized Systems Used in Clinical Trials", draft of June 18, 1997, U.S. Department of Heath and Human Services, Food and Drug Administration.
http://www.datafax.com - provides fax (rather than digital data) solutions for forms
http://www.oracle.com - provides back end databases as well as the Oracle Clinical application for managing aspects of clinical studies
http://www.sas.com - provides powerful statistical analysis solutions
http://www.sagargroup.com - a contract research organization and information technology provider
New Drug Development: A Regulatory Overview, 4th ed., M. Mathieu  (Parexel, 1997)
Systems Analysis and Design, 3rd ed., Kenneth E. Kendall and Julie E. Kendall (New Jersey: Prentice Hall, 1995).

You are visitor number: Hit Counter