Remote Data Capture in Clinical Trials - PDF by ndb36858

VIEWS: 60 PAGES: 5

Remote Data Capture in Clinical Trials document sample

More Info
									                                                                                                     1




Data Capture Technology:
From Remote Data Entry to Direct Data Capture.

Andy Hyde


Remote data entry (RDE), in the conventional       its ability to check data as it is being entered. If
sense has failed. At best, RDE - using             an error or omission is detected, the person
computers with electronic forms installed to       doing the data entry can be warned so that
collect clinical trials data - functions           corrective action can be taken immediately.
acceptably in a limited number of trials for a     This obviously has a positive effect on the
limited number of data types. Current              quality of the data returned to the sponsor. The
estimates predict that no more than 5% of all      second greatest benefit is the time saved in
trials will ever use RDE as the primary method     transferring the data into the sponsor's central
of data collection 1. But technology cannot be     clinical trials database. When this is planned
ignored.                                           and tested early in the process, data can flow
   Considering the possibilities, we could use     smoothly from one system to the other,
our increasing knowledge of computing's            reducing the chance of introducing errors into
strengths and seek a way to capitalize on          the data. Other benefits include control over
technology's potential. More and more data-        additional comments and improved legibility,
providing systems are based on computers or        which can be problems in paper CRFs; and
technology, making it possible to connect a        reduce paper usage, a valuable contribution in
data collection computer to a data provider        these days of environmental consciousness.
together and thus to use direct data capture          Yet, after more than 10 years in
(DDC) to collect data.                             development, current EDC methods, primarily
   A parallel development may have an even         remote data entry, have been implemented into
greater effect on data collection: the             only an estimated 5% of clinical studies - a
development of hospital systems for collecting     percent that is not increasing.
patient data from its current equipment and its       If there is to be a future for RDE, it will be
legacy systems, the electronic patient record      found in an Internet based solution. Several
(EPR) or electronic health record (EHR). Once      pilot trials are being conducted to examine the
a hospital or other health care provider has an    feasibility of using Internet technology. Early
EHR or EPR in a database, it becomes possible      results show that using ActiveX or Java
for clinical trials personnel to extract from it   Applets is currently too slow. Using some kind
the information they require. This can             of server-based data checking with an
minimize duplication of data entry - a growing     extremely thin client - it has been called an
problem at hospitals and investigator sites -      "anorexic" client - works better. These
and prevent the data quality problems caused       solutions will, however, still suffer from the
by repeated entry and transcription.               same inadequacies of their forebears. All data
                                                   must be "pre-prepared" and manually entered
Remote Data Entry                                  into the data entry system.
                                                      The expectations placed on RDE systems
Electronic Data Capture (EDC) refers to any
                                                   were shortened study duration, reduced errors
method of electronically capturing data for a
                                                   and resource savings. Only one expectation has
clinical trial. In practice, RDE - using basic
                                                   been consistently demonstrated; a reduction in
personal computer (PC) technologies - has
                                                   error rates in the data returned from the
been used almost exclusively for electronic
                                                   Investigator. Many parts of the process other
data capture at clinical trial sites. A typical
                                                   than RDE affect trial duration. And, so far,
RDE implementation includes a desktop or
                                                   little evidence has emerged to demonstrate
laptop PC with an electronic form application
                                                   reduced use of resources - either human or
installed. An electronic case report form
                                                   financial.
(eCRF) replaces the paper CRF.
   The eCRF has several advantages over the
traditional paper CRF. The greatest of these is
                                                                                                    2


Direct Data Capture                                    "Smart Cards" are another future source of
                                                     subject information. These optical storage
RDE is just one way of collecting data into a
                                                     devices, the size of a credit card, can store all
computer. As implemented, RDE requires
                                                     of a person's health-related information along
several steps between source and data entry.
                                                     with security measures, such as photo
But advances in technology are rapidly
                                                     identification. The notes taken at each visit can
providing a realistic alternative, direct data
                                                     be stored, as can information about drugs
capture (DDC). Direct data capture is base on
                                                     prescribed. These cards can hold up to 4
the concept of putting source data directly data
                                                     megabytes of compressed information - the
collecting computer system.
                                                     equivalent of 2000 A4 or letter-sized pages -
   With RDE all source data is put into one
                                                     enough for a lifetime for most people.
standard format before it is entered into the
                                                       DDC can reduce the number of errors even
computer. With DDC, data can arrive in the
                                                     further than RDE. It can save the Investigator
computer in many different formats; then the
                                                     time and effort. It does however require the
computer can manipulate the data to meet one
                                                     development of a multitude of different
standard for transfer into the central repository.
                                                     interfaces to the different data providing
   DDC is not a new concept. Many
                                                     sources.
laboratories already use the method, and many
                                                       A future source of data for DDC is becoming
sponsors     receive     electronic     laboratory
                                                     available through technological developments
information that can be imported directly into
                                                     in the health care industry. Health care
their clinical databases. The samples are
                                                     institutions and the pharmaceutical industry
analyzed by a machine which outputs its
                                                     have been facing the same EDC problems. At
results in computer form, ready to be re-
                                                     many health care sites, the solution is the
formatted      into a standard layout and
                                                     electronic patient record (EPR) and the
transmitted electronically to the sponsor. Once
                                                     electronic health record (EHR).
at the sponsor site, the data can be up-loaded
into the central repository.
   A rapidly growing number of clinical trials       EPRs and EHRs
data sources are becoming available                  EPRs and EHRs are an exciting development
electronically, for example. Equipment for           for information management - a development
various aspects of physical examinations -           driven mainly by managed heath care
ECG, EEG, magnetic resonance imaging                 organizations (MCOs) in the United States and
(MRI) and nuclear imaging machines for               by the movement to reduce costs for U.S.
example - can all create digital output. Image       insurance companies and government-funded
data can be stored in a standard format called       health care in Europe. By combining all the
DICOM (Digital Imaging Communications in             information on patient care for hundreds or
Medicine) - a standard protocol for                  thousands of patients receiving a given drug or
transmitting medical images, waveforms, and          treatment with treatment outcomes, third-party
ancillary information. Information about the         payers can identify expensive drugs or
technical parameters of those machine is stored      treatments with poor outcomes and use that
in the image file as part of this standard and       cost/benefit information to reduce costs. To
can readily be extracted and transferred into        combine information on so many patients
the clinical data store. With tradition paper        requires a computerized system, hence the
based CRFs, these kind of technical data have        need for EPRs.
been associated with a high error rate and              There is a slight difference between the EPR
entering them into an eCRF is time consuming.        and the EHR. The EPR assumes that the
   DDC is also applicable to non-electronic          person is a patient and is therefore in need of
sources. A large amount of data provided by          some examination or treatment. The EHR has
the research subjects themselves. In RDE-            broader aims: to register the heath information
based studies these data are elicited through        of individuals whether or not they are current
verbal questioning, written into the CRF and         patients in need of care. This then extends the
finally entered into the eCRF. With increased        information to smoking habits, dietary
computer literacy and ownership, however,            information, exercise routines and more,
many subjects can fill in information at a           including data collected outside the health care
terminal whilst they wait to see a doctor - or       institution.
even while with the doctor. In the future,              For healthcare institution (hospitals for
providing information via the Internet before a      example) - where implementation of
visit will become an acceptable data collection      Information Technology has often grown in
technique.                                           uncontrolled ways - developing an EPR is a
                                                     major undertaking. Each department and
                                                     function autonomously selected a computer
                                                                                                     3


system suited to its own needs without
forseeing thatit would one day need to                  Translation & Transfer via HL7
communicate with other systems.
  Creating an EPR can be done in two ways.
                                                         HL7 is a messaging system in which
The first is to replace all of a health care
                                                      messages are structured according to a pre-
provider's systems with one new distributed
                                                      defined format and sent from one system to
system based on one platform and one
                                                      another. The sending system needs to know
application. But that is a major undertaking,
                                                      only how to convert its data into an HL7
one for which few health care institutions are
                                                      message and the receiving system needs to
setup     physically,     organizationally  and
                                                      know how to extract those data. Message are
politically. This approach may be impossible.
                                                      sent when triggered. The Admissions.
  An alternative is to retain all the systems
                                                      Discharges and Transfers (ADT) system for
currently in place and get them to "talk" to
                                                      example has a trigger that activates whenever a
each other. This is the solution that many
                                                      new patient is admitted to a hospital. The
vendors choose. A standard was developed for
                                                      trigger says electronically, "send the name,
communicating between the systems called
                                                      insurance company, address and other
HL7 (Health Level 7, in which the 7 indicates
                                                      particulars to all other systems." An example
that it operates at level 7 of the Open Systems
                                                      of an HL7 message is:
Interface (OSI) 7 layer model for networking).
The HL7 standard is approved by ANSI
                                                       MSH|^~\&|ADT1|MCM|LABADT|MCM|1
(American National Standards Institute) and is
                                                      98808181126|SECURITY|ADT^A01|MS
based on consensus of its members. It is              G00001|P|2.3|<cr>
possible to add any new data collection system         EVN|A01|198808181123||<cr>
into the EPR as long as it can communicate in          PID|||PATID1234^5^M11||HYDE^AN
HL7.                                                  DREW^W^Jr||19610615|M||C|1200 N
                                                      ELM
Technology choices for EHRs                           STREET^^GREENSBORO^NC^27401-
The technology choices made by the                    1020|GL|(919)379-1212|(919)271-
healthcare industry should indicate what is           3434||S||
acceptable for an investigator's routines and a        PATID12345001^2^M10|123456789|
hospital's functions. Choices thus far have been      987654^NC|<cr>
pragmatic - to avoid changing the daily                NK1|HYDE^KIRSTEN^G|WIFE||||||N
practice of the caregiver. Data must be               K^NEXT OF KIN<cr>
collected at the point of care and require no          PV1|1|I|2000^2012^01||||004777
subsequent transcription. In this way, the            ^JAMES^SIDNEY^J.|||SUR||||ADM|A
information is immediately available to other         0|<cr>
caregivers who may require it. Some of the
technologies available are voice recognition,           The message above translates to: Patient
laser disk storage and structured remote data         Andrew W. Hyde, Jr., was admitted on August
entry.                                                18, 1988 at 11:23 a.m. by doctor Sidney J.
   Voice recognition is currently high on the         James (#004777) for surgery (SUR). He has
list. Much of the information collected on a          been assigned to room 2012, bed 01 on nursing
patient visit is recorded to tape, then sent to the   unit 2000. The message was sent from system
medical records department for transcription.         ADT1 at the MCM site to system LABADT,
"Off-line" dictation of this kind can still be        also at the MCM site, on the same date as the
done with digital tape recorders and voice            admission took place, but three minutes after
recognition software, or a doctor can dictate         admission.
directly into a desktop computer. With training
(and an American accent in a quiet
environment) voice recognition works well. A
host of medical dictionary add-ons make it            a laser disk. The record can be regenerated by
both possible and highly effective.                   reading the data back into the template.
   Laser disk storage. Another technology is            Structured data entry, such as that for
computer output to laser disk (COLD)                  RDE, is far down the list of choices. It requires
whereby free text is typed into a predefined          specialized software and it takes longer to
word processor template document, then                collect the same data than voice recognition,
"printed" via an interface that can interpret the     laser disk storage or paper case record forms
structure of the form and the text in it. The text    (CRFs). The greatest chance for the adoption
and the template are split and the data are           of structured data entry is increases in access
written to a mass storage device - in this case,      to Internet technology, which is becoming a
                                                                                                         4


more common way to make information                       A number of machines are being connected
available at the point-of-care. Health-care            up to a TCP/IP (Transmission Control
institutions also use Intranets inside the             Protocol/Internet Protocol) communications
institution and Extranets between them. Charts,        systems so they can send data into a central,
images and text based information are often            technologically homologous storage. A
presented on web pages. Structured data entry          modern MRI machine is a good example; it
at the point of care is therefore feasible.            can be hooked up to the Internet to send
                                                       pictures wherever they are needed. For this
From RDE to DDC                                        kind of data transfer to occur there must be
                                                       standards similar to the HL7 standard for
To move from RDE to DDC it is important to
                                                       communicating between imaging systems.
throw away the old CRF model. We must
                                                       DICOM is such a standard. It defines
adjust ourselves to capturing source data
                                                       structures and protocols for transmission and
directly into a computer. The current CRF is
                                                       storage so that any DICOM compatible
likely to be used in only a limited number of
                                                       computer can read the pictures.
situations - and probably have another name.
                                                          If both the pharmaceutical industry and the
Computers have been around long enough for
                                                       health care industry move to DDC, the result
clinical trials and health care professionals to
                                                       could be a parallel effort that captures patient
understand and be comfortable with the issues
                                                       data into two different systems. That
of validation, backup and security. With
                                                       introduces the potential for inconsistency and
careful planning, these issues present few
                                                       an unnecessary amount of extra work - work
problems.
                                                       that will be particularly noticeable to the
   The changes required to move forward from
                                                       investigator and other site staff responsible for
remote data entry are as great as, if not greater
                                                       collecting the data. To put the final piece of the
than, the changes required in moving from
                                                       puzzle in place and achieve the ideal
paper CRFs to electronic CRFs. (The eCRF is
                                                       information flow, would be ideal to first
still a complete record in one place of the
                                                       capture all the data into one system, then
subject's progress through the trial). With the
                                                       selectively export it to the other systems.
change to DDC, information will be more
patient-based with addition information
collected specific to the patient's involvement        The future of DDC
as a subject in the clinical trial. If a concept of    If one assumes that a person is in the health
a CRF remains, it will be of a virtual CRF.            care institution primarily as a patient and
Today people in document management can                secondarily as a subject in a clinical trial, then
now work with virtual documents in the same            it is logical to see the data collected first in the
way. Several people write a document at the            EPR, then exported to the sponsor as the
same time. Then, when it needs to be viewed            patient-subject completes study participation.
as a complete document, it is "stitched"               But that process raises a number of issues, not
together by the document management system.            least of which are patient confidentiality, data
   When a virtual CRF model is accepted by             quality issues and data coding standardization.
the clinical trials industry, it will be possible to      Confidentiality. One suggestion from the
move forward by identifying all the possible           health care industry to ensure confidentiality
data sources and plan the best way to capture          and control is to make the data extraction from
data directly. This is as much a technological         the EPR protocol driven. Clinical protocols
challenge as a process planning challenge.             must drive the extraction utility so that only
Data may well exist in several machines'               the data required, and no more, are transferred.
permanent storage or on a variety of external             Data quality. Data will be owned by the
storage devices such as CD-ROM, diskette,              trial site - a requirement of ICH/GCP - and the
DVD (the digital versatile disk, originally the        quality of the information will be the site's
digital video disk), MO (magneto optical) disk,        responsibility. Any error that the sponsor
and others. Data stored in all of these media          discovers must first be corrected in the EPR
must be captured          into one system (the         system before being "re-extracted" to the
complexity of the task provides an interesting         sponsor. Here, the pharmaceutical industry's
challenge      for     technology      enthusiasts).   experience with front end error checking from
Reducing the number of storage options or              RDE systems and the use of standard operating
capturing the data over a network are                  procedures (SOPs) to ensure data quality can
alternative solutions. Capturing over a network        be used to address quality issues in health care
involves using a communication standard such           data systems.
as HL7 or DICOM and then using the target                 Standardization. Standardization issues
machine to store the data in a consistent              must be addressed to enable consistent storage
physical format.                                       and communication of the required data. The
                                                                                                    5


latest version of the HL7 standard, v3.0,           electronic data capture conferences over the
contains messages for clinical trials. The          next 10 years.
messages include such things as; subject              The change will neither be easy nor fast, but
registration and study completion of a patient,     the benefits are there to be reaped, particularly
trial phase information, treatment schedules,       for the pharmaceutical. The use of DDC, with
sponsor information, randomization codes,           the extraction of data from EPR systems,
subject consent information, evaluable status.      should produce cost and time savings. The
   Two other standards for treatment and drug       health care industry will save because the new
prescription information are widely used in the     systems will remove much of the technological
health care industry: the ICD9 (International       duplication now beginning to emerge in the
Classification of Diseases, 9th revision) and       investigator's daily routine. Clinical trials
SNOWMED (Systemized Nomenclature of                 professionals will gain because data collection
Medicine) dictionaries. HMOs use them to            and error correction will be simplified during
assign standard codes to items that will be         and after clinical trials.
charged to insurance companies and also as the        The cost effectiveness of the DDC will
basis for determining the treatment/outcome         depend upon the level of standardization
for cost/benefit analysis. The pharmaceutical       achieved. Without a high level of
industry uses the World Health Organization         standardization,           each        individual
(WHO) drug and adverse event dictionary and         implementation will cost too much. One way
the new MedDRA (Medical terminology for             to spread the cost of implementing a DDC
Drug Regulatory Affairs, a resource similar to      program is to reuse study centers, thus
ICD10) dictionary for recording diagnosis,          spreading the cost across several studies.
adverse events and reactions, and elements of         Although RDE has not fulfilled new
product characteristics summaries.                  technology's potential for data capture, new
   The Health Information Management and            technology is, and will be, an increasingly
Systems       Society     (HIMSS)      and    the   important component of data collection. To
Radiological Society of North America               avoid duplication and technological overload
(RSNA) are working together in a joint              for the investigators upon whom clinical trials
initiative, Integrating the Healthcare Enterprise   rely so heavily, it is important to identify the
(IHE), to co-ordinate and promote standards.        technological synergies that can simplify the
This is a good opportunity to promote one           work of health care clinical trials professionals.
standard across those two disciplines.
                                                    References
From vision to reality
Before the pharmaceutical industry and the          1.   A.W., Hyde, 1998 New Technology
health care industry can capitalize on                   Systems being tried in the collection of
technology's potential, vendors of both EPRs             clinical trials data: a user centred
and EDC systems must co-operate. Only such               comparison using HCI methods. Authors
co-operation can ensure a standard and co-               unpublished Masters Degree dissertation.
ordinated approach.                                      (available electronically by request to the
   In this bright future of co-operation, one            author)
final move must be made: accepting the
Internet as a safe and effective means of
transferring clinical trials data. The data can     Andy Hyde, MSc (CCI) (Open), information
flow seamlessly across the Internet from the        management and technology consultant,
EPR system to the sponsor. But don't look for       Nycomed Amersham Imaging, Nycoveien 1-2,
this to happen tomorrow!                            PO Box 4220, 0401 Oslo, Norway, +47
   Even with the development of Web-enabled         23185312, fax +47 23186003, e-mail:
RDE systems, we're unlikely to see much             ahy@nycomed.com.
innovation or trial penetration in RDE. It is
more likely that DDC will be the theme for

								
To top