VA Research: Protections for Human Subjects Need to Be Strengthened
(Letter Report, 09/28/2000, GAO/HEHS-00-155).

Pursuant to a congressional request, GAO reviewed rights and welfare of
veterans who volunteer to participate in research at Department of
Veterans Affairs (VA) and the effectiveness of its human subject
protection system, focusing on: (1) VA's implementation of human subject
protections; (2) whether weaknesses exist in VA's system for protecting
human subjects; and (3) VA's actions to improve human subject
protections at those sites affected by sanctions applied by regulatory
agencies and throughout VA's health care system.

GAO noted that: (1) VA has adopted a system of protections for human
research subjects, but GAO found substantial problems with its
implementation of these protections; (2) medical centers GAO visited did
not comply with all regulations to protect the rights and welfare of
research participants; (3) among problems GAO observed were failures to
provide adequate information to subjects before they participated in
research, inadequate reviews of proposed and ongoing research,
insufficient staff and space for review boards, and incomplete
documentation of review board activities; (4) GAO found relatively few
problems at some sites that had stronger systems to protect human
subjects, but GAO observed multiple problems at other sites; (5)
although the results of GAO's visits to medical centers cannot be
projected to VA as a whole, the extent of the problems GAO found
strongly indicates that human subject protections at VA need to be
strengthened; (6) three specific weaknesses have compromised VA's
ability to protect human subjects in research; (7) VA headquarters has
not provided medical center research staff with adequate guidance about
human subject protections and thus has not ensured that research staff
have all the information they need to protect the rights and welfare of
human subjects; (8) insufficient monitoring and oversight of local human
subject protections have permitted noncompliance with regulations to go
undetected and uncorrected; (9) VA has not ensured that funds needed for
human subject protections are allocated for that purpose at the medical
centers, with officials at some medical centers reporting that they did
not have sufficient resources to accomplish their mandated
responsibilities; (10) to VA's credit, substantial corrective actions
have been implemented at three medical centers in response to sanctions
by regulatory agencies taken against their human research programs, but
VA's systemwide efforts at improving protections have been slow to
develop; (11) medical centers affected by sanctions have taken numerous
steps to improve human subject protections; and (12) VA has, however,
been slow to take action to identify any systemwide deficiencies and
obtain necessary information about the human subject protection systems
at its medical centers.

--------------------------- Indexing Terms -----------------------------

 REPORTNUM:  HEHS-00-155
     TITLE:  VA Research: Protections for Human Subjects Need to Be
	     Strengthened
      DATE:  09/28/2000
   SUBJECT:  Veterans
	     Medical research
	     Health research programs
	     Safety regulation
	     Safety standards
	     Noncompliance
	     Informed consent (medical law)
	     Research program management

******************************************************************
** This file contains an ASCII representation of the text of a  **
** GAO Testimony.                                               **
**                                                              **
** No attempt has been made to display graphic images, although **
** figure captions are reproduced.  Tables are included, but    **
** may not resemble those in the printed version.               **
**                                                              **
** Please see the PDF (Portable Document Format) file, when     **
** available, for a complete electronic file of the printed     **
** document's contents.                                         **
**                                                              **
******************************************************************
GAO/HEHS-00-155

Appendix I: Scope and Methodology

46

Appendix II: Comments From the Department of Veterans Affairs

50

Appendix III: GAO Contact and Staff Acknowledgments

55

56

Table 1: Characteristics of Sampled VA Medical Centers 47

Figure 1: Human Research Project Review and Approval Process
at VA Medical Centers 11

Figure 2: Noncompliance With VA Regulations at Eight Sites 12

Figure 3: Elements of Informed Consent Required by VA
Regulations 14

Figure 4: VA's Regulatory Criteria for IRB Approval of Research
Projects 20

FDA Food and Drug Administration

HHS Department of Health and Human Services

IRB institutional review board

NIH National Institutes of Health

OHRP Office for Human Research Protections

OPRR Office for Protection from Research Risks

ORCA Office of Research Compliance and Assurance

ORD Office of Research and Development

VA Department of Veterans Affairs

Health, Education, and
Human Services Division

B-283287

September 28, 2000

The Honorable Terry Everett
Chairman
The Honorable Corrine Brown
Ranking Democratic Member
Subcommittee on Oversight and Investigations
Committee on Veterans' Affairs
House of Representatives

The Honorable Cliff Stearns
Chairman
The Honorable Luis V. Gutierrez
Ranking Democratic Member
Subcommittee on Health
Committee on Veterans' Affairs
House of Representatives

Every year thousands of veterans volunteer to participate in biomedical or
behavioral research projects under the auspices of the Department of
Veterans Affairs (VA). They do so for different reasons, including to
improve the medical conditions of themselves and others, advance science,
and serve their country. But research is not without risks. VA studies, like
other federally funded research programs, are governed by regulations
designed to minimize risks and protect the rights and welfare of research
participants. VA must ensure that veterans who agree to become subjects in
VA research are given accurate and understandable information about
procedures, risks, and benefits so that they can make informed decisions
about volunteering. However, veterans who rely on VA for their health care
often do so because they are unable to afford private health care and may
fear jeopardizing their care if they do not agree to participate in
research. VA has a special responsibility, therefore, to ensure that
veterans are not unduly influenced in their decision to participate in
research projects and that their rights and welfare are protected if they
volunteer to participate.

Concerns about VA's protection of its human research subjects came to
national attention in March 1999. At that time, all human research was
suspended at the West Los Angeles VA Medical Center after officials there
failed to correct long-standing problems with its system for protecting
human subjects. Concerns about human subject protections have not been
limited to VA. Public attention has recently been focused on concerns about
the safety of research subjects in other public and private research
institutions. Department of Health and Human Services (HHS) investigations
of violations of human subject protections have led to the suspension of
research activities at several universities and hospitals. Furthermore, the
HHS Office of Inspector General has raised concerns about the adequacy of
current human subject protections nationwide in a series of reports. These
reports note that protections for research subjects are threatened by recent
changes in the research environment that have included a heightened industry
role in sponsoring research, proliferation of multicenter trials, increased
number and complexity of research proposals, increased demand by patients to
participate in clinical trials, and changes in health care delivery systems
affecting research programs.

Since the suspension of human research at the West Los Angeles VA Medical
Center in March 1999, four additional VA medical centers have been affected
by sanctions applied by regulatory agencies against their affiliated
universities. Concerned about the rights and welfare of veterans who
volunteer to participate in research at VA and the effectiveness of its
human subject protection system, you asked us to (1) assess VA's
implementation of human subject protections, (2) identify whether weaknesses
exist in VA's system for protecting human subjects, and (3) assess VA's
actions to improve human subject protections at those sites affected by
sanctions applied by regulatory agencies and throughout VA's health care
system.

To address these objectives, we reviewed VA regulations and guidance and
interviewed headquarters officials responsible for human subject research
activities. We visited eight VA medical centers to review their policies,
procedures, and practices for protecting human subjects. We selected these
medical centers to reflect major differences in VA research programs,
including the number of studies they conduct with human subjects and the
institutions responsible for operating the local institutional review board
(IRB), the committee tasked with reviewing a research study's protections
for human subjects. At these medical centers, we examined IRB activities to
determine whether appropriate protections for human subjects were in place.
We also visited three other VA medical centers affected by recent
restrictions of their human research programs and reviewed the changes
implemented there in response to HHS regulators. We talked with senior
research officials at two additional VA medical centers similarly affected
by sanctions applied by regulatory agencies. Because research programs at
five of the VA facilities we visited involved university affiliates, we also
reviewed documents and interviewed officials responsible for human subject
protection issues at those institutions. We performed our work from June
1999 through August 2000 in accordance with generally accepted government
auditing standards. See app. I for a detailed discussion of our scope and
methodology.

VA has adopted a system of protections for human research subjects, but we
found substantial problems with its implementation of these protections.
Medical centers we visited did not comply with all regulations to protect
the rights and welfare of research participants. Among the problems we
observed were failures to provide adequate information to subjects before
they participated in research, inadequate reviews of proposed and ongoing
research, insufficient staff and space for review boards, and incomplete
documentation of review board activities. We found relatively few problems
at some sites that had stronger systems to protect human subjects, but we
observed multiple problems at other sites. Although the results of our
visits to medical centers cannot be projected to VA as a whole, the extent
of the problems we found strongly indicates that human subject protections
at VA need to be strengthened.

Three specific weaknesses have compromised VA's ability to protect human
subjects in research. First, VA headquarters has not provided medical center
research staff with adequate guidance about human subject protections and
thus has not ensured that research staff have all the information they need
to protect the rights and welfare of human subjects. Second, insufficient
monitoring and oversight of local human subject protections have permitted
noncompliance with regulations to go undetected and uncorrected. Third, VA
has not ensured that funds needed for human subject protections are
allocated for that purpose at the medical centers, with officials at some
medical centers reporting that they did not have sufficient resources to
accomplish their mandated responsibilities.

To VA's credit, substantial corrective actions have been implemented at
three medical centers in response to sanctions by regulatory agencies taken
against their human research programs, but VA's systemwide efforts at
improving protections have been slow to develop. Medical centers affected by
sanctions have taken numerous steps to improve human subject protections,
for example, by hiring and training IRB staff and developing written
procedures for their IRB operations. Despite some difficulties, these
medical centers have made progress, and each has resumed human research
activities. VA has, however, been slow to take action to identify any
systemwide deficiencies and obtain necessary information about the human
subject protection systems at its medical centers. VA has made promising
steps, for example, by establishing the Office of Research Compliance and
Assurance to monitor human subject protections at individual medical centers
and across the system and developing a system to use an external
organization to accredit IRBs. However, it is too soon to determine whether
they will fulfill their objectives.

In light of these problems, we make recommendations to the Acting Secretary
of Veterans Affairs to strengthen VA's protections of the rights and welfare
of human subjects by providing staff training and resources and taking other
steps to ensure that medical centers and their IRBs comply with all
applicable human subject protection regulations. VA concurred with our
recommendations.

Conducting research is one of VA's core missions. VA researchers have been
involved in a variety of important advances in medical research, including
development of the cardiac pacemaker, kidney transplant technology,
prosthetic devices, and drug treatments for high blood pressure and
schizophrenia. For fiscal year 2000, Congress appropriated $321 million for
VA's research programs, which support a wide range of human, animal, and
basic science studies. VA uses a competitive funding process in which its
Office of Research and Development (ORD) allocates about $296 million of
these funds to VA researchers, with awards based on scientific merit and
potential contribution to knowledge of issues of particular concern to VA.
VA allocates most of the remainder to indirect costs of research, which
includes support for the human subjects protection system.

Besides the appropriation for research, VA allocates funds from its medical
care appropriation to support the research infrastructure at medical centers
such as laboratory facilities and investigator salaries. In fiscal year
2000, this allocation amounted to $343 million. VA researchers receive
additional grants and contracts from other federal agencies such as the
National Institutes of Health (NIH), research foundations, and private
industry sponsors, including pharmaceutical companies. In fiscal year 1999,
these additional funds amounted to approximately $481 million. Nonprofit
research foundations linked to VA medical centers control some of these
non-VA research funds.1 In fiscal year 2000, biomedical or behavioral
research involving human subjects is being conducted at about 70 percent of
VA medical centers.

VA is responsible for ensuring that all human research it conducts or
supports meets the requirements of VA regulations, regardless of whether
that research is funded by VA, the subjects are veterans, or the studies are
conducted on VA grounds. Responsibility for administration and oversight of
the research program has rested primarily with ORD. Recently, VA created the
Office of Research Compliance and Assurance (ORCA), which has been charged
with advising the Under Secretary for Health on all matters affecting the
integrity of research protections for humans and animals, promoting the
ethical conduct of research, and investigating allegations of research
improprieties.

Some VA research is also subject to oversight by two HHS components. The
Food and Drug Administration (FDA) is responsible for protecting the rights
of human subjects enrolled in research with products it regulates--drugs,
medical devices, biologics, foods, and cosmetics. Research that involves
human subjects and is funded by HHS is subject to oversight by its Office
for Human Research Protections (OHRP).2 HHS requires institutions conducting
human research with HHS funds to file a document with OHRP that indicates a
commitment to comply with federal regulations. This document, called an
assurance, may cover a single study (a single project assurance), or it may
allow the institution to conduct multiple studies (a multiple project
assurance).3 When an institution files a multiple project assurance with
OHRP, all federally funded research involving human subjects at that
institution must comply with HHS regulations. Both FDA and OHRP have the
authority to monitor those studies conducted under their jurisdiction, and
each can take action against investigators, IRBs, or institutions that fail
to comply with applicable regulations.

Research with human subjects conducted at VA facilities is governed by
regulations designed to protect their rights and welfare. These regulations
establish minimum standards for the conduct and review of research to ensure
that research involving human subjects is conducted in accordance with the
three ethical principles outlined by the National Commission for the
Protection of Human Subjects of Biomedical and Behavioral Research.4 First,
the principle of respect for persons requires acknowledgement of individual
autonomy, and conversely, the need to protect those with diminished
autonomy. In practice, this principle requires that subjects give informed
consent to participate in research; that is, they must be given sufficient
information about a study?including its purpose, procedures, risks, and
benefits?to decide whether to participate. They must also understand this
information, and their consent must be voluntary. Second, the principle of
beneficence requires that the expected benefits of research to the
individual or to society outweigh its anticipated risks. Third, the
principle of justice requires fair subject selection procedures, so that
both the benefits and the burdens of research are distributed across a
number of individuals in a just manner.

In 1981, in response to the National Commission, both HHS and FDA
promulgated revised regulations for the protection of human subjects.
Seventeen federal departments and agencies, including HHS and VA, have
adopted the core of HHS regulations.5 FDA's regulations are slightly
different from those adopted by HHS and VA.

To safeguard the rights of subjects and promote ethical research, these
federal regulations create a system in which the responsibility for the
protection of human subjects is assigned to three groups.

ï¿½ Investigators are responsible for conducting their research in accordance
with applicable federal regulations and for ensuring that legally effective
consent is obtained from each subject or his or her legally authorized
representative.

ï¿½ Institutions are responsible for establishing oversight mechanisms for
research including establishing local committees known as institutional
review boards (IRB), which are responsible for reviewing research proposals
before studies are initiated and after they are under way to help ensure
that research is conducted in accordance with the three principles described
above.

ï¿½ Agencies, including VA, are responsible for ensuring that their IRBs
comply with applicable federal regulations and that they have sufficient
space and staff to accomplish their obligations.

VA requires each of its medical centers that engages in research with human
subjects to establish its own IRB6 or secure the services of an IRB at an
affiliated university. As of August 2000, approximately 40 percent of the
medical centers conducting research with human subjects relied on an IRB at
an affiliated university. The IRB sends its recommendations to the VA
medical center's research and development committee, which is responsible
for maintaining standards of scientific quality, laboratory safety, and the
safety of human and animal subjects. The research and development committee
is charged with reviewing each study's budget; assessing the availability of
needed space, personnel, equipment, and supplies; and determining the effect
of the planned research on the investigator's other responsibilities,
including the provision of clinical services. The committee can disapprove a
study; however, VA regulations prevent the research and development
committee (or any other institutional official or body) from overturning an
IRB decision to disapprove a study.

A VA investigator who wants to conduct research with human subjects must
develop a research plan (called a protocol), supporting documents, and a
consent form. The consent form is designed to provide potential subjects
with sufficient information about the study, including its procedures,
risks, and benefits, to allow the subject to make an informed decision about
whether to participate in the study (see fig. 1). The investigator then
submits these materials for review. The study is not to be initiated until
both the IRB and the research and development committee have approved it,
and these committees may insist on changes to the protocol or consent form.
Once approval has been given, VA regulations prohibit any unapproved changes
to the study's procedures, unless doing so is absolutely necessary to ensure
the safety of a subject. If an investigator wants to alter some aspect of
the study, then the IRB must review and approve an amendment or modification
to the protocol. In a process known as continuing review, each study is to
be re-reviewed at least once per year, and more frequently if the degree of
risk warrants it.

Centers

We found variation across medical centers and their affiliated universities
in the implementation of VA regulations and policies involving protections
for human subjects. At the eight sites we visited, we found noncompliance
with VA regulations in four areas: (1) informed consent; (2) IRB review; (3)
IRB membership, staff, and space; and (4) IRB documentation. The problems we
identified are similar to problems that OHRP noted in letters to
universities and hospitals it has found to be out of compliance with federal
regulations. As shown in fig. 2, some sites we visited had more problems
than did others.

aWe reviewed from 14 to 20 IRB-approved consent forms at each site. We
reviewed a total of 138 forms.

bWe compared consent forms signed by subjects to IRB-approved consent forms
for 17 to 20 studies at each of four sites. We compared forms for a total of
73 studies.

Of the sites we visited, those with the most extensive violations of VA
regulations relied on VA-run IRBs. We identified fewer problems at the IRBs
in our sample that were run by universities. In particular, we observed
fewer problems with IRB membership, staff, space, and documentation at
university-run IRBs than at VA-run IRBs. University-run IRBs were also more
likely to conduct thorough and timely continuing reviews than VA-run IRBs.
University-run IRBs we visited were not without problems, however. We found
that some IRB-approved consent forms at each site omitted required
information and some investigators used nonapproved consent forms.

We found problems with the content or use of informed consent forms at all
of the medical centers we visited. We found that some informed consent
documents that had been approved for use by IRBs provided incomplete or
unclear information. In addition, we found some studies in which the
investigators used nonapproved consent forms when enrolling subjects. We
also found one instance in which research was conducted without consent.
Informed consent is a primary ethical requirement of research with human
subjects and reflects the principle of respect for persons. The ability of
competent subjects to make informed decisions about whether to participate
in research and the ability of legally authorized representatives to protect
those who are unable to provide consent because they are incapacitated are
undermined when IRBs fail to ensure that all required information is
included in consent forms or when investigators fail to obtain consent using
approved procedures.

Information in IRB-Approved Consent Forms

We found that 60 percent of the 138 IRB-approved consent forms that we
randomly sampled from lists of active projects provided incomplete or
unclear information about required elements of informed consent. (Fig. 3
lists the elements of informed consent required by VA regulations.) Each IRB
we visited approved some consent forms that contained incomplete
information. For example, IRB-approved consent forms did not

ï¿½ indicate that blood would be drawn in a study on the effects of exposure
to Agent Orange,

ï¿½ mention possible risks of a biopsy in a study designed to test a treatment
for esophageal cancer,

ï¿½ describe alternative treatment options in a study comparing two drug
treatments for schizophrenia, and

ï¿½ indicate who would have access to data obtained during a study on
treatment for cirrhosis of the liver.

Source: VA Regulations (38 C.F.R. section 16.116).

Of the 84 IRB-approved consent forms we identified that omitted required
elements or provided incomplete information, almost half did so for two or
more required elements. For example, the consent form for a study of
treatments to reduce the recurrence of melanoma did not provide clear
information about the duration of the study, nor did it state whom to
contact for information about research subjects' rights. Participants were
also told that data would continue to be obtained from their medical records
even if they withdrew from the study. Thus, the consent document for this
study provided incomplete information about two required elements and
appeared to negate the subject's right to withdraw from the study at any
time. Moreover, this consent form might have created undue influence because
it inappropriately suggested that the subjects' own physician endorsed the
potential benefits to the subject of participating in this study. Because
the participants in this study are randomly assigned to receive either an
unproven treatment or no treatment, the physician would have no way of
knowing whether participation would benefit the subject.

VA regulations allow an IRB to approve a consent procedure that alters or
omits one or more of the required elements of consent if it finds and
documents certain conditions. We were unable to find such documentation in
the cases we reviewed. Moreover, 37 of the IRB-approved consent forms that
omitted or provided incomplete information about a required element were for
studies that involved investigational drugs or devices. Thus, both VA and
FDA regulations had to be met, and when informed consent is required, FDA
regulations do not permit IRBs to alter or omit any required elements of
informed consent.

The information that was omitted most frequently--in about 15 percent of
forms?was the person to whom subjects should direct questions about their
rights as research subjects. This information, which is required by
regulations, is not included in the standard template for informed consent
that VA policy requires investigators to use.

Sites varied in the number of IRB-approved forms that provided incomplete
information and the number of incomplete or absent elements in approved
forms. The percent of approved consent forms with incomplete information
ranged from 78 to 100 percent of our sample at the four sites with the
greatest number of these problems. Moreover, forms from these four sites
often provided incomplete descriptions of two or more required elements of
informed consent. As many as four elements of informed consent were missing
or incomplete in IRB-approved forms at these sites. At the two sites where
we found the fewest problems, about three-fourths of our sample of approved
consent forms were problem-free, and multiple problems in the same form were
rare.

In addition to information required by VA regulations, VA policy also
requires that informed consent forms indicate that VA will provide free
medical treatment for research-related injuries.7 We found that about 30
percent of the IRB-approved consent documents we reviewed did not include
this statement. The absence of this statement varied by site. (These data
are not included in fig. 2, which presents noncompliance with VA
regulations.) The majority of forms we sampled at two university-run IRBs
did not include this information, and one VA-run IRB included it only about
half the time. In contrast, the forms at the other university-run IRBs and
at the four other VA-run IRBs almost always included it.

The requirement for informed consent was waived for eight of the projects we
reviewed, and in each case, our review indicated that the study qualified
for the waiver. According to VA regulations, certain categories of
research?for example, studies of existing data that cannot be linked,
directly or indirectly, to specific individuals?do not require informed
consent or IRB approval. VA regulations also allow for a waiver of informed
consent in some research that is not eligible for an exemption from IRB
review, provided that the IRB determines that certain conditions apply.

Investigator Noncompliance With Consent Requirements

Although all the consent forms we obtained from investigators indicated that
consent to participate in research had been obtained, we found that
investigators did not always obtain consent appropriately. In this review of
consent forms, we found 18 studies in which the investigators used
nonapproved consent forms when enrolling subjects. We also separately
identified one instance in which research was conducted without consent.

We asked investigators at each site to show us signed consent forms from a
randomly selected sample of their subjects. We examined 540 such consent
forms, all of which had the signature of a subject or a surrogate. In
addition to determining that investigators were able to produce these signed
consent forms, at four sites we also compared these signed forms with
consent forms that IRBs had approved for use in these studies. We found that
investigators had used nonapproved consent forms with one or more subjects
in 18 of the 73 studies we examined. A total of 33 of 292 subjects had
signed nonapproved consent forms. The extent of this problem varied by site.
We found that one or more subjects had signed a nonapproved form in 12 to 33
percent of the studies we examined at these four sites. Some of the
nonapproved forms that were signed by subjects omitted key information that
had been included in the IRB-approved version of the consent form. For
example, the nonapproved form that had been signed by all four subjects
enrolled in a study on treatments for lymphoma did not mention that the
study would involve multiple bone marrow biopsies, the possible risks of
those biopsies, or possible side effects of two drugs used in the
study?information that was included in the IRB-approved consent form.

We identified one instance in which research procedures were performed
without consent in the projects in our sample. In this instance, a patient
who had not given consent was subjected to an esophageal biopsy for research
purposes. This biopsy, which was not reported to the IRB, occurred in
conjunction with a biopsy performed for diagnostic purposes in November
1997.

We also found that investigators or their staff had not fully complied with
requirements for obtaining consent in three other studies in our sample. In
each of these, subjects had consented and steps were implemented to address
the problem.

ï¿½ In October 1998, an investigator learned that a subject with schizophrenia
did not understand his right to withdraw from research at any time. Upon
discovering this, the investigator fired the person who had obtained the
subject's consent, withdrew the subject from the study, and reported the
incident to the IRB.

ï¿½ In May 1997, FDA discovered that the consent form signed by subjects in a
study of an investigational device to facilitate walking among paraplegics
had not included all the necessary information about their participation.
The problem was reported to the IRB, the consent form was rewritten, and
three previously enrolled subjects were given a revised form and a chance to
withdraw from participation.

ï¿½ In July 1997, an investigator realized that he did not have IRB approval
for the protocol and consent form that had been used for 73
subjects?including schizophrenics, their family members, and health care
providers?who had completed a questionnaire to assess decision-making. The
investigator reported the situation to the IRB, which required that subjects
be given a revised approved consent form.

We found one other problem in subject enrollment procedures used by an
investigator, although in this case VA regulations were not involved. One
subject who was incapacitated as a result of dementia was enrolled?contrary
to VA policy?in a noninvasive study of abdominal aneurysms. Although the
subject's surrogate had provided consent,8 VA policy establishes protections
for incapacitated subjects by prohibiting their enrollment in research that
can be conducted with competent subjects. We encountered eight other cases
in which surrogates enrolled incapacitated subjects in research, but we were
unable to determine whether these cases were in accordance with VA's policy.

We found that five of the sites we visited did not implement certain
required procedures for IRB review of research. For example, we found that
studies at two sites were not reviewed by all necessary IRB members and four
IRBs did not ensure timely or thorough continuing review of ongoing
research.

Initial IRB Review--
Required Approval During Convened IRB Meetings

We found that two IRBs did not comply with VA regulations that research must
be approved during properly convened meetings, either because meetings were
held without a quorum or because the IRB chair improperly approved a
high-risk study outside an IRB meeting. With the exception of certain
categories of research involving minimal risk to subjects, VA regulations
require IRBs to review research at convened meetings attended by a quorum,
defined as a majority of members that includes at least one member whose
primary concerns are in nonscientific areas. These regulations establish
criteria for IRB meeting quorums to ensure that decisions about the
protection of human participants in research reflect the consideration of
diverse perspectives on research, including the views of scientists and
nonscientists with a range of experience and expertise. These protections
are undermined when initial review is not conducted in accordance with these
requirements.

Four of seven meetings held by one VA-run IRB between January 1998 and
August 1999 were held without a quorum. As a result, 17 studies were
initiated without legitimate IRB approval, including studies on new drug
treatments for unstable coronary symptoms and pneumonia. We examined four to
six sets of minutes from IRB meetings held at the other seven IRBs we
visited and found that a quorum was present at each.

We found one other instance in which requirements for approval of research
at convened IRB meetings were violated. A university-run IRB considered a
high-risk drug study for cardiac patients and determined that re-review
would be necessary after the investigator addressed several concerns. IRB
minutes stated that because the drug company sponsoring the research would
have rejected their site if a time deadline were not met, the IRB chair
approved the study before the IRB reconsidered it. Although there are
circumstances under which an IRB chair can approve a study, in all such
cases the research must have been found to pose only minimal risk to
subjects. In this instance, the IRB had determined that the study posed a
high degree of risk.

On the other hand, our sample also included 16 other studies that met
criteria for approval outside a convened IRB meeting. VA regulations allow
such a procedure (called expedited review) for studies that pose only
minimal risk to subjects and that fall into one of several categories of
research. Under expedited review procedures, the IRB chair, or one or more
experienced IRB members designated by the chair, are authorized to approve
research. For example, IRB approval was expedited for a study on the effects
of a weight loss program in which subjects would attend informational
sessions about diet and weight loss and have their weight and health
monitored using routine, minimal-risk procedures.

Initial Review?Sufficiency of Written Information From Investigators

We found that the IRBs we visited differed in the sufficiency of the written
information they asked investigators to provide about human subject
protections prior to review. VA regulations identify eight criteria that
IRBs must assess before approving research (see fig. 4). Although VA
regulations do not specify the information IRBs must review to assess these
criteria upon initial review, much of the information can only be provided
by investigators. Because offsite study sponsors often prepare the consent
forms and protocols used in multisite studies, IRBs must have sufficient
information to assess whether the local investigator can properly implement
human subject protections.

We found that information in IRB files did not always address all the
criteria that must be satisfied for an IRB to approve a study. Of the sites
we visited, only one university-run IRB routinely requested detailed
information from local investigators about each criterion in its application
forms. For example, two IRBs did not routinely ask local investigators any
questions about risks or about plans for monitoring the safety of subjects.

Source: VA Regulations (38 C.F.R. section 16.111).

Similarly, IRBs differed in the information they had from investigators
about special protections for subjects who are likely to be vulnerable to
coercion or undue influence. VA regulations require that IRBs ensure that
additional safeguards are in place to protect the rights and welfare of such
subjects; however, the regulations do not specify the nature of such
safeguards. We analyzed project files for 27 studies designed to address
issues involving psychiatric conditions that can be associated with a
diminished capacity for decision-making?psychoses, mood disorders, and
organic mental disorders such as dementia. We found that the investigator
had included information about additional safeguards in applications for IRB
approval in only about half of these studies. For example, we reviewed from
two to six files for projects involving potentially vulnerable subjects at
each site and found references to additional protections in most of the
relevant project files we sampled at four sites. In contrast, at two other
locations no such documentation was evident in any of the IRB files we
reviewed for projects involving subjects with psychiatric disorders that
could affect decision-making.

Some sites have implemented procedures that afford special protections for
some such subjects. Examples follow.

ï¿½ Subjects at one medical center who are recruited for psychiatric research
and whose mental illness can affect decision-making are typically tested for
their comprehension of central consent issues before enrollment in a study.

ï¿½ At another medical center, seriously mentally ill subjects who participate
in studies involving a risk that their symptoms might worsen are monitored
by a physician who is independent of the research and who is assigned
responsibility for deciding whether the subject should remain in the study
or be withdrawn.

ï¿½ Alzheimer's researchers at a third site have established research
registries for potential subjects, who were still able to give consent, and
their caregivers. By enrolling, subjects agree to allow medical information
to be entered into a data bank and to be contacted about future studies. By
agreeing to be contacted, however, potential subjects have not consented to
participate in future studies. Because these potential subjects are
recruited for future studies through registries, the risk of undue influence
that occurs when physicians recruit their patients is minimized. Moreover,
rules for these registries limit the number of researchers who may contact
each person, ensure that potential subjects are recruited only for studies
for which they are in fact eligible, and allow registry managers to conduct
follow-up surveys to ensure that members of the registries are satisfied
with the way researchers treat them.

Continuing Review

We found that three VA-run IRBs did not meet VA's regulatory requirement
that each study must be re-reviewed at intervals not to exceed 1 year.9
Regular re-review of a project and associated reports of problems allows an
IRB to assess the ratio of risks to benefits on the basis of data obtained
since the study began and to ensure that subjects are appropriately informed
of those risks and benefits. We examined the dates of continuing review for
73 projects at 6 sites that had received initial approval more than 1 year
before our visit. Of these projects, 54 (74 percent) had been reviewed on
time within the past year. The median delay for the 19 projects that were
not re-reviewed on time was about 1 month. At one VA-run site, only one of
the nine projects we reviewed that were more than 1 year old had been
re-reviewed on time. At another VA-run site, about half of the necessary
continuing reviews from our sample were conducted within 1 year, but delays
of up to 14 months occurred in the other half. The three university-run IRBs
we visited achieved high rates of timely continuing review.

Four VA-run IRBs we visited reviewed insufficient information when
conducting continuing review. OHRP has stated that compliance with
regulatory requirements for continuing review entails, at a minimum, IRB
review of

ï¿½ the study protocol and any amendments;

ï¿½ the current consent form;

ï¿½ the number of subjects who have been enrolled; and

ï¿½ information relevant to risks, including adverse events, unanticipated
problems involving risks to the subject or others, withdrawal of subjects
from the study, complaints about the study, and a summary of any recent
information relevant to risk assessment.

Only half of the IRBs we visited required the investigator to submit the
most recent version of the consent document or asked about subjects who have
withdrawn (or been withdrawn) from the study. All eight IRBs required
reports of the number of subjects who had participated and adverse events.

IRB staff told us that reports of adverse events are difficult for IRBs to
handle. Regulations require investigators to report to the IRB unanticipated
problems involving risks to subjects, and IRBs must review adverse events
reported by all sites where the study is being conducted. The concerns we
heard on our site visits were similar to those described in several recent
reports on difficulties that IRBs nationwide face when handling large
numbers of adverse event reports in the absence of key information necessary
for their interpretation.10 For example, reports of adverse events from drug
studies do not indicate whether the subject who experienced the adverse
event had received an experimental drug or a different treatment, such as a
placebo. Regulatory bodies such as FDA and OHRP and research sponsors such
as the National Cancer Institute have recently argued that adverse event
reports from studies involving many subjects are often best handled by
special committees called data and safety monitoring boards. These boards
are typically established by research sponsors and include statisticians and
other scientists who analyze data collected during the course of a clinical
trial to detect risks to subjects. A few of the IRBs we visited were
attempting to develop systems to track adverse events.

Even when a data and safety monitoring board has been established to analyze
adverse event reports associated with a study, it is not required to report
its findings to IRBs. In VA these boards, referred to as data monitoring
boards, analyze only those adverse events reported in multicenter studies
funded by VA through a program called Cooperative Studies. If results
indicate that a study protocol or consent form must be modified, reports are
released by the coordinating center for that cooperative study. It sends
such reports to investigators and to the associate chiefs of staff for
research and development at participating medical centers, with instructions
to share the information with IRBs. Reports are not submitted to IRBs
directly. Similarly, VA's policy manual does not require that reports from
data and safety monitoring boards associated with non-VA-funded research be
submitted to its IRBs or medical centers. VA's policy manual also does not
require investigators or IRBs to ascertain whether a data and safety
monitoring board has been established for studies in which its investigators
participate.

IRBs at the eight facilities we visited met certain membership requirements,
but two did not ensure that their members had no potential conflicts of
interest. We also found problems involving the number of IRB staff or IRB
space at five facilities. VA regulations require that IRBs have sufficient
administrative staff and space to review research and preserve the
confidentiality of files.

VA regulations for IRB membership include requirements that IRBs have at
least five members and must include a scientist, a nonscientist, and at
least one person who is not otherwise affiliated with the institution.
(Individual members may fulfill more than one criterion.) We checked IRB
membership rosters from the eight facilities we visited and found that all
met these requirements. In addition, VA regulations state that if the IRB
regularly reviews research involving a vulnerable category of subjects, then
consideration should be given to including at least one member who has
experience working with that group. Each of the eight IRBs we visited
included someone from the institution's psychiatry, psychology, or other
mental health department, allowing access to specialized expertise with
regard to the potential vulnerabilities of mentally ill subjects.

We also found that each of the university-run IRBs we visited had members
who were on staff at the affiliated VA medical center. Inclusion of VA staff
helps fulfill VA's regulatory requirement that IRBs have knowledge of the
local research institution, including the scope of research activities,
types of subjects likely to be involved, and the size and complexity of the
institution. Officials at the medical centers we visited that relied on the
IRBs of university affiliates reported that the larger academic community of
the university offered advantages for IRB membership, including a broader
range of expertise and reduced potential for conflicts of interest because
IRB members would be less likely to be research colleagues of investigators.
In addition, because all VA investigators at these three medical centers
also held faculty appointments at the university, investigators did not need
to apply for IRB approval from both the university and VA. Officials at some
of the medical centers that operated their own IRBs reported that the
advantages of doing so included maintaining greater control over the
research review process and the increased likelihood that the IRB would know
particular investigators and veteran subjects.

We found that two VA-run IRBs did not ensure that their members had no
potential conflicts of interest. VA regulations state that no IRB may have a
member participate in an IRB initial or continuing review of any project in
which that member has a conflict of interest. Although we found that
investigators who were IRB members appropriately abstained or recused
themselves from voting on their projects, two IRBs had, as a voting member,
the associate chief of staff for research and development for their medical
centers. The duties of a VA medical center's associate chief of staff for
research and development include helping local investigators obtain
intramural or extramural research funds. As noted by OHRP, such
institutional officials thus have a potential conflict of interest in
conducting IRB reviews. These two officials told us, however, that they
believed their objectivity as IRB members was not compromised by their other
responsibilities.

Officials at four of the VA-run IRBs told us that they did not have adequate
staff to support IRB operations, as required by VA regulations. IRB
administrative staff provide crucial services such as reviewing applications
for completeness, corresponding with investigators, and maintaining IRB
records. In addition, some administrative staff serve on IRBs as experts on
regulatory issues. The VA-run IRBs we visited typically had one or two IRB
staff members who often had other responsibilities. For example, at one of
these sites, where a single staff person worked part-time for an IRB that
reviews 200 to 300 projects annually, the IRB chair reported that IRB
activities, such as suggesting revisions to consent forms, were curtailed
due to insufficient staff support. In May 2000, VA headquarters distributed
preliminary estimates for the number of administrative IRB staff that a
medical center should have. This guidance noted that staffing levels would
vary with the breadth and complexity of the research program. ORD officials
acknowledged that these benchmarks are a first approximation in an effort to
identify appropriate staffing levels.

In addition to staff, IRBs must have secure, private areas for the review
and discussion of confidential materials. IRBs also need office space for
the IRB chair and administrative staff, secure file storage, and computer
support. We found that IRB administrative staff at three sites--two of them
VA-run IRBs?lacked sufficient space to conduct their work or store all IRB
documents. For example, we observed IRB file folders stacked loosely on top
of file cabinets and on floors at one of these sites.

Six of the eight IRBs we visited did not maintain all the records required
by VA regulations. Inadequate documentation does not, in itself, place
subjects at risk. However, records of actions, deliberations, and procedures
can help identify problems and corrective actions. Thus, documentary
failures prevent appropriate monitoring and oversight activities.

We found inadequate documentation in IRB files for about 9 percent of the
ongoing projects we reviewed. For example, some files failed to include
copies of all correspondence regarding IRB actions between the IRB and
investigators, or copies of all approved consent forms. VA regulations
require IRBs to retain these documents for at least 3 years after a study is
terminated. Required documents were missing from one or more IRB files at
five of the eight sites we visited.

VA regulations require each facility to maintain written procedures that it
will follow for conducting initial and continuing review, reporting IRB
findings and actions to investigators and appropriate officials, and
determining when special steps are necessary to monitor ongoing projects.
Our review indicated wide differences between facilities in the adequacy of
these documents. One VA-run facility has written procedures regarding
criteria for exemption from IRB review and for use of expedited review
procedures that are not in accordance with VA regulations. In addition, one
medical center had been cited by the FDA for failure to have adequate
written procedures in June 1999. The center agreed to have them in place by
August 1999 but did not do so until December 1999. The written procedures
available from three other VA-run IRBs did not include required descriptions
of procedures for conducting project review, determining when additional
monitoring of projects is necessary, or responding to investigator
noncompliance. In contrast, the written procedures of the three
university-run IRBs included all required procedures.

We found one instance in which failure to have required written policies
resulted in a further violation of VA regulations. Specifically, the
previously discussed esophageal biopsy, which was conducted without consent,
was not reported to the IRB or OHRP as required. VA regulations require
institutions to ensure that "serious or continuing noncompliance"11 by
investigators is reported to the IRB. A similar report must be filed with
OHRP if the institution has an HHS-approved assurance, as did the medical
center involved. The Associate Chief of Staff for Research and Development
told us that he did not report the event to the IRB or OHRP because he
followed the procedures for handling scientific misconduct outlined in VA's
policy manual. Nothing in the IRB's project files for that investigator
indicated a finding or report of noncompliance, imposition of any special
restrictions or conditions for future research, or suspension or termination
of research.

We found that some IRB minutes did not comply with VA regulations, which
require the minutes to include a record of actions, the basis for requiring
changes in or disapproving research, and a written summary of discussions of
controverted issues and their resolution. At each site, we reviewed from
four to seven sets of minutes from IRB meetings held from December 1997
through October 1999. IRB actions were almost always clearly recorded in the
minutes we examined at each site. Minutes from six facilities routinely
included written summaries of discussions and reasons for actions. Two
VA-run IRBs, however, rarely included substantive discussions of these
matters in their minutes.12

Facilities also varied in their compliance with VA regulations about
recording votes by IRB members during project review. The regulations state
that minutes of IRB meetings must indicate the number of members voting for
and against and the number of those abstaining. Two VA-run IRBs typically
recorded votes as unanimous, and minutes from one other VA-run IRB recorded
some votes as "approved," without specifying vote totals. Without exact
numbers, the presence of a majority of IRB members required during each vote
cannot be confirmed. The voting records in minutes from the remaining IRBs
we visited were generally in compliance with regulations. However, in one
set of minutes from one site, we found that the total number of votes cast
for each decision consistently exceeded the number of members listed in
attendance.

We identified three specific weaknesses in VA's system for protecting human
subjects: not ensuring that research staff have appropriate guidance,
insufficient monitoring and oversight activity, and not ensuring that the
necessary funds for human subject protections are provided. These weaknesses
indicate that human subject protection issues have not historically received
adequate attention from VA headquarters.

VA headquarters has not provided the guidance necessary to ensure that its
medical center staff are adequately informed about requirements for the
protection of human research subjects. We found that VA did not develop a
systemwide educational program, ensure that each of its facilities had an
appropriate training program in place, or provide guidance about training to
its facilities. We also found problems with the guidance VA provides about
procedures for handling informed consent records. Efforts to protect the
rights and welfare of human subjects are undermined when research staff have
not been given clear, comprehensive guidance about human subject
protections.

VA headquarters officials told us that VA did not have a systemwide
educational program devoted to human subject protection issues and that more
training is needed. We found that three of the medical centers we visited
had no educational program for IRB members, IRB staff, or investigators.
From its October 1999 survey of VA field management, VA headquarters
research officials learned that 12 of 22 Veterans Integrated Service
Networks13 did not have an adequate plan for the ongoing education of IRB
members, IRB administrative staff, or investigators about the regulatory
requirements for protecting human subjects. In particular, medical centers
with small research programs identified difficulties in establishing
educational programs. Those facilities that had programs often reported that
their university affiliates ran the training programs. A need for increased
educational guidance from headquarters was one of the most commonly
identified issues regarding human subject protections in the survey. OHRP
and HHS's Office of Inspector General have stressed that educational
programs are critical to ensuring that IRBs comply with regulations and are
able to assess the acceptability of research proposals in light of those
regulations and to ensuring that investigators understand their
responsibilities to protect human subjects.

On the other hand, two VA-run IRBs and the three university-run IRBs we
visited have implemented their own educational programs for both
investigators and IRB members and staff, generally without guidance from
headquarters. These programs included training new IRB members, devoting a
portion of IRB meetings to discussion of issues involving the protection of
human subjects, having some IRB members and staff attend national
conferences about IRB operations, and instituting a certification program
for investigators. Although we did not evaluate the adequacy of these
programs, one of these sites, a university affiliate, developed an
educational program that has been cited by HHS's Office of Inspector General
as a best practice for training in human subject protection issues.

In addition to finding that VA did not have a systemwide educational
program, we found problems with VA guidance for documenting consent to
participate in research. VA's policy manual includes two requirements that
go beyond its regulations for the protection of human subjects: (1) the
original signed consent form is to be placed in the subject's medical record
and (2) investigators are to use a standard template developed by VA to
obtain consent.

A VA official in ORD told us that the purpose of requiring the placement of
signed research consent forms in medical records is to ensure that treating
professionals are aware of relevant medical information. He acknowledged,
however, that consent forms in medical records are not always readily
accessible to treatment staff because they may be housed in old volumes of
medical records maintained in storage areas. He also noted that medical
records personnel at some VA medical centers have discarded consent forms
rather than filing them. Our findings confirmed this. We were unable to
locate consent forms in 20 percent of 187 medical records we reviewed at 7
of the 8 medical centers we visited. The remaining medical center we visited
recently developed a system for scanning signed consent documents into its
electronic medical records. However, these consent forms were not located in
a part of the electronic record that would be routinely accessed by treating
personnel.

Some medical center research staff suggested that placing a synopsis of each
study in a prominent place within subjects' medical records would ensure
that treating professionals know about relevant research participation, thus
minimizing risks to subjects. We observed such a strategy at the Denver VA
Medical Center, where a special flag in each subject's electronic medical
record links the reader to a brief summary of the study and to any
investigational drugs involved. VA has not implemented a systemwide
procedure for indicating research involvement in electronic medical records.

Another area of concern is VA's standard template for informed consent. This
template includes space for investigators to enter study-specific
information and exact language for requirements common to all consent forms.
VA's policy manual requires all VA investigators to use this form. We
identified several problems with this template.

ï¿½ The template does not reflect the regulatory requirement that a contact be
provided for subjects to call with questions about their rights as research
participants.

ï¿½ For studies conducted at both VA and non-VA locations, use of the VA
template created problems. In these cases, adherence to VA's policy requires
development and IRB approval of two consent forms--one based on VA's
template and one for the other location. Failure to use an appropriate
IRB-approved consent form in these dual-form studies was the reason subjects
signed nonapproved forms in 10 of the 33 cases previously discussed.

ï¿½ VA has not provided clear guidance about the role of a witness to the
consent process. Under VA regulations, a witness signature is needed only
when the elements of informed consent have been presented orally. We found
only 1 study in our sample of 146 in which consent was obtained orally.
However, we found that 405 of the 540 signed consent forms we examined had
been signed by a witness. OHRP guidance indicates that a witness to a
subject's consent to participate in research may be appropriate when aspects
of the study create concerns about the enrollment process. In such cases, an
independent witness can provide a valuable check on the consent process to
certify, for example, that key information was properly conveyed and that
subjects were not unduly coerced into participation. On the other hand, such
a witness can represent an unnecessary intrusion into a potential subject's
privacy. VA's consent template includes a line for the signature of a
witness, without specifying who may serve as a witness, what the witness is
attesting to, or the circumstances under which the witness is needed.
Similarly, VA's policy manual lacks guidance about who should serve as a
witness or what that person's role is.

We found that VA did not have an effective system for monitoring protections
of human subjects. Several instances follow.

ï¿½ VA headquarters and affected medical centers were generally unaware of
regulatory investigations and impending actions by OHRP or FDA against
university-run IRBs until after the regulatory sanctions were applied. VA
was unable to ensure that FDA could notify VA of planned inspections and
provide copies of post-inspection correspondence because VA was unable to
provide FDA with a list of its university-run IRBs until July 2000. VA did
not have a complete list of those medical centers that used their own IRBs,
relied on a university-run IRB, or were covered by an OHRP assurance until
July 2000.

ï¿½ Until OHRP's regulatory action against the West Los Angeles VA Medical
Center, VA was unaware that each of its facilities was required to provide a
written assurance that it will comply with all federal regulations regarding
the protection of human subjects. Written assurances facilitate proper
oversight by ensuring documentation of core agreements between VA
headquarters and IRBs. They also can provide evidence of knowledge of the
regulations governing human subject protections and demonstrate an
institution's commitment to those protections. When VA subsequently obtained
these assurances, it did not require medical centers to submit local written
procedures for implementing human subject protections, as the regulations
required. Review of written procedures can indicate gaps or errors in
required local policies and procedures.

ï¿½ VA headquarters has not provided medical centers with guidance in ensuring
access to minutes or other key information when they arrange for the
services of a university-run IRB. As a result, one medical center we visited
did not have access to the minutes of its university-run IRB, and two
medical centers affected by regulatory sanctions against their affiliated
universities had not monitored IRB minutes to assess compliance with
regulations.

Furthermore, we found that VA headquarters and medical centers we visited
did not effectively monitor investigators and their studies. Specifically,
only one of the eight medical centers we visited checked whether
investigators provided subjects with the correct IRB-approved consent form.
That medical center recently began checking one signed consent form from
each study as part of its continuing review. In addition, the files of one
university-run IRB we visited did not correctly identify which researchers
at the VA medical center were responsible for the studies the IRB had
approved because the medical center required that department chairs rather
than researchers be listed as principal investigators.

Responsibility for funding human subject protections at medical centers is
diffused across several decisionmakers, each of whom may also have competing
priorities for the same funds. As a result, no one official is responsible
for ensuring that medical center research programs have the resources
necessary to support IRB operations and provide training in human subject
protections. Although VA has not determined the funding amounts needed for
human subject protection activities at the medical centers, research
officials at five of the eight medical centers we visited told us that they
had insufficient funds to ensure adequate operation of their human subject
protection systems.

We found that medical centers typically relied on several sources of funds
to support the indirect costs of research, which include human subject
protection activities. These sources included VA's research appropriation,
VA's medical care appropriation, and non-VA research sponsors such as NIH or
pharmaceutical companies. Different decisionmakers control the funds
potentially available to a medical center from these sources.

ï¿½ The medical center's associate chief of staff for research and development
controls the portion of the research appropriation targeted for the indirect
costs of research.

ï¿½ The medical center's director controls the portion of the medical care
appropriation allocated for indirect costs of research.

ï¿½ Funds from non-VA research sponsors are generally held by a medical
center's nonprofit research foundation and are controlled by its board of
directors, which has discretion over their use.

As a result, responsibility for ensuring that human subject protections are
adequately funded at each medical center is diffused across several
decisionmakers. In addition, the decisionmakers at some of the medical
centers we visited told us that they did not allocate additional funds for
human subject protection activities because they had to consider those needs
against the competing priorities of research support and medical care
delivery. Headquarters research officials confirmed that these
organizational tensions have created a situation in which there is no clear
focus of responsibility for funding human subject protection activities at
medical centers.

One of the indirect costs of operating an IRB is the time spent by IRB
chairs and members meeting their IRB responsibilities. Headquarters research
officials told us that providing release time for IRB chairs and members has
been a long-standing problem. VA staff at the medical centers we visited
conduct their IRB activity as a collateral duty. We were told that the time
commitment for members, and particularly for IRB chairs, is significant.
Chairs and members spend time reviewing protocols before meetings,
corresponding with investigators, attending IRB meetings, and preparing and
reviewing documentation. We were told that the lack of release time made it
difficult to recruit and retain IRB chairs and members. We found one
instance in which a university paid VA to subsidize the costs of covering
the emergency room duties of a VA physician who chaired an IRB that VA used.
In another instance, a research official at one medical center told us that
IRB meetings are held in the evening and that the nonprofit foundation pays
IRB members. This arrangement allows members to fulfill their primary VA
obligations during the day without the collateral responsibility of serving
on the IRB.

Research officials at five of the eight medical centers we visited reported
that they had insufficient funds to ensure adequate operation of their human
subject protection systems. Of particular concern, officials told us, was
that lack of funds prevented hiring and training staff. Officials from some
medical centers also told us that their nonprofit research foundations
recognized that the level of VA funding for IRB operations was inadequate,
and therefore contributed varying amounts of funds for specific local needs,
such as training investigators in human subject protections or hiring IRB
staff. For example, one nonprofit contributed $25,000 in fiscal year 2000 to
support investigator training in human subject protections. Some VA
nonprofit foundations and universities are charging private industry
sponsors a fee for IRB review of their projects to help support IRB
operations. However, headquarters research officials told us that VA has not
determined the funding amounts needed for human subject protection
activities at the medical centers. They said that such a determination is
necessary for planning funding levels and ensuring that human subject
protection activities are appropriately funded.

Systemwide Focus Slow to Develop

Substantial corrective actions have been implemented at three medical
centers in response to sanctions by regulatory agencies against their human
research programs. These steps represent progress in meeting the
requirements imposed by regulators and VA management, and each of the
facilities, despite some difficulties, has resumed human research
activities. VA has, however, been slow to identify systemwide deficiencies
and to obtain information needed to step up oversight of human subject
protection systems at its medical centers. Nonetheless, VA's recent
responses, such as establishment of the Office of Research Compliance and
Assurance (ORCA) to monitor human subject protections at individual medical
centers and across the system, are promising.

The three medical centers and their affiliated universities we visited that
had actions taken against them by regulators--West Los Angeles, Chicago
Westside, and Denver--have made progress in implementing substantial changes
to their human subject protection systems. Their written procedures appear
to be in compliance with regulations, and their staffing levels seem
reasonable for the workload.

These medical centers and their affiliated universities, along with two
others, had been affected by serious regulatory sanctions. Regulators found
numerous problems at these institutions, including failure to obtain
informed consent, failure to conduct adequate and timely continuing review
of research, and failure to have adequate written IRB policies and
procedures. OHRP deactivated West Los Angeles VA Medical Center's multiple
project assurance with HHS on March 22, 1999. It restricted the assurance
held by the University of Illinois at Chicago, which served as the IRB of
record for Chicago Westside VA Medical Center on August 27, 1999. On
September 13, 1999, FDA suspended certain research projects at a consortium
of six Colorado research institutions, including the Denver VA Medical
Center. The University of Colorado, the location of the consortium's IRB,
suspended research with human subjects at all six sites in response to a
letter from OHRP dated September 22, 1999, which raised concerns about IRB
noncompliance with regulations. On December 17, 1999, OHRP restricted the
multiple project assurance with Virginia Commonwealth University, which had
been the IRB of record for the Richmond VA Medical Center. FDA had issued a
warning letter to the university several months earlier about the IRB
operations. On January 19, 2000, OHRP restricted the multiple project
assurance with the University of Alabama at Birmingham, which was the IRB of
record for the Birmingham VA Medical Center.

There were three immediate responses in West Los Angeles, Chicago, and
Denver to the sanctions imposed by regulatory agencies: a suspension of
enrollment of new subjects in almost all research projects; an assessment of
the appropriateness of the continued participation of previously enrolled
subjects; and a determination by VA headquarters and affiliated universities
of actions needed to improve human subject protection programs at each site.
Each medical center or affiliated university that we visited then made
extensive changes to its human subject protection system. These changes
involved reconstituting IRBs; increasing the number of IRB administrative
staff; training IRB members, staff, and investigators in the principles and
procedures of human subject protection; creating or extensively revising IRB
procedures; increasing working space for IRB operations; creating new
databases for tracking protocols through the review process; re-reviewing
projects; and resuming research activities. As of February 2000, all
projects at the West Los Angeles VA Medical Center had been re-reviewed by
an IRB. As of June 2000, all projects for the Chicago Westside VA Medical
Center had been submitted to university-run IRBs for re-review, and as of
July 2000, all projects had been re-reviewed for the Denver VA Medical
Center. The Denver VA Medical Center's IRB has been informed by OHRP and FDA
that as of June 2000, its corrective actions are appropriate. On July 18,
2000, OHRP removed the restriction on the University of Illinois at Chicago
stating that the university has developed and implemented an improved system
for the protection of human subjects in research and has adequately
completed all required actions.

Responses varied across sites, however, because of differing
responsibilities for IRB operations and site-specific problems that needed
to be addressed. For example, at the West Los Angeles VA Medical Center,
which operated its own IRB, VA headquarters and medical center officials
made extensive changes in research personnel responsible for human subject
protections. From April 1999 to the time of our visit in March 2000, about
50 employees had been rotated through the program with a few assigned
full-time to support research and development and IRB operations. The
university affiliated with the Chicago Westside VA Medical Center hired a
nationally known expert in human subject protections to lead a comprehensive
restructuring of its IRB operations.

We identified two issues of concern at the West Los Angeles VA Medical
Center. First, VA's authorization of a resumption of IRB operations at West
Los Angeles on April 19, 1999--less than 1 month after OHRP's deactivation
of its multiple project assurance--was premature. At that time, the medical
center still lacked approved, written procedures for operation. Such
procedures are required by regulations. It also was relying on untrained
administrative staff to assist the newly formed IRBs. Furthermore, VA's
investigators had not been trained in human subject protection issues.

Our second issue of concern is that officials at the West Los Angeles VA
Medical Center were particularly slow to respond to OHRP's requirements. In
its 1999 letter deactivating the medical center's multiple project
assurance, OHRP noted the medical center's continued lack of responsiveness
to issues raised by OHRP over a 5-year period. For example, in 1994, OHRP
required that the medical center establish a data and safety monitoring
board to oversee studies involving subjects with severe psychiatric
disorders. It took until February 2000 for medical center officials to
approve standard operating procedures for the data and safety monitoring
board and to hire its staff.14 In another instance, OHRP cited the medical
center in 1995 for a lack of adequate written procedures for human subject
protections. However, it took the medical center until February 2000 to
develop and approve these procedures. Similarly, in 1995, OHRP strongly
recommended that medical center officials develop an ongoing training
program for investigators. Medical center officials told us they plan to
begin such training in September 2000.

At the Chicago Westside VA Medical Center, we found that, in permitting the
continued participation of previously enrolled subjects in some projects, VA
and the university-run IRBs did not ensure that continuing review
requirements were met for these projects. When we raised this issue with
officials during our February 2000 visit, they acknowledged this lack of
oversight. They have since required investigators for these projects to
submit materials for continuing review.

We found that the Chicago Westside VA Medical Center did not play an active
role in assisting its university-run IRBs to improve its human subject
protection system. The medical center organizational chart for research and
development did not show any linkage with the three university IRBs. The
medical center had only one representative among the 18 members of the
biomedical IRB and one on the 17-member combined biomedical-behavioral IRB.
There were no VA representatives on the third IRB, an IRB that reviewed
behavioral studies because, as officials told us, VA conducted few such
studies. At the time of our visit to the medical center--over 5 months after
the OHRP action--the medical center had done little to improve its
communication with the IRBs despite the recommendation to do so made by the
VA headquarters site visit team in September 1999. Although one local VA
research official participated on a university committee charged with
prioritizing studies for re-review and made suggestions to modify the IRB
form used by investigators to submit protocols for review, the medical
center had not established a mechanism for routine contact with and
monitoring of the IRBs. In addition, the medical center was unaware of VA
protocols being submitted for IRB review, IRB actions to approve or
disapprove continuation of studies, and serious adverse events that could
affect veterans who were subjects of research. At the time of our visit, the
medical center was unable to provide us with reliable data on which
investigators had been trained by the university in human subject protection
regulations and issues. Furthermore, as of July 2000, the medical center had
not responded to a May 2000 request from the university for comments on
their new IRB procedures manual.

In contrast, the Denver VA Medical Center established mechanisms to enhance
communication between the research and development program and its three
university-run IRBs by having regular meetings and increasing the number of
VA personnel on the IRBs. As of June 2000, the chair of one of the
university-run IRBs and the co-chair of another were VA employees. Five
other VA employees served as members of the IRBs. Medical center personnel
were working closely with their counterparts in the university to design a
database that would allow VA research officials access to VA project
information at the university-run IRBs.

When the IRBs at their affiliated universities faced sanctions by regulatory
agencies, officials at the Richmond and Birmingham medical centers chose to
establish their own IRBs. They told us they did so to increase their control
over the research review process. These officials told us they each created
an IRB, developed written procedures, trained IRB members, and resumed their
research programs after re-reviewing their projects. In addition, the
Birmingham VA Medical Center has trained investigators and IRB staff, and
the Richmond VA Medical Center has trained research staff.

VA has been slow to recognize and address systemwide deficiencies in its
human subject protection activities. Although OHRP identified problems with
human subject protections at the West Los Angeles VA Medical Center in 1994,
VA did not have a plan to address systemwide concerns involving research
until July 1998. VA did not begin to implement systemwide changes until
after OHRP took regulatory action against the medical center in March 1999.
VA's initial responses to regulators' actions affecting the West Los Angeles
VA Medical Center and other medical centers were crisis-driven and
site-specific. Specifically, headquarters formed teams that conducted site
visits to determine actions needed at the affected medical centers.
Headquarters monitored corrective actions at the medical centers primarily
through an exchange of reports and correspondence.

In July 1998, VA developed a plan to reorganize its field research
operations. This plan addressed a variety of research concerns including the
involvement of human subjects and the ethical conduct of studies. Only
recently, however, has VA headquarters begun to implement systemwide changes
to improve its human subject protections. Its steps have included providing
information to investigators and research staff, obtaining information about
medical centers' research programs, and making organizational changes to
enhance monitoring and oversight of research involving human subjects. These
steps have been slowly implemented, but they provide a promising foundation
for improvements to protections for human subjects in VA research.

Providing Information to Investigators and Local Research Staff

VA headquarters officials have taken several steps to provide information to
VA investigators and local research staff about human subject protections.
The initial information provided by ORD described issues at affected medical
centers. It was not until October 1999 that ORD provided medical centers
with specific actions that could be helpful in strengthening their human
subject research programs.

Starting with its May 1999 bimonthly conference call with associate chiefs
of staff for research and development, ORD began discussing human subject
protection issues in light of the March 1999 OHRP action against the West
Los Angeles VA Medical Center. Also in May 1999, they began to plan a series
of educational programs for investigators, IRB members, research
administrators, and medical center directors focused on human subject
protection issues. In October 1999, ORD held a nationwide videoconference in
which OHRP and VA research officials discussed human subject protection
issues and answered questions from VA staff. Also in October 1999, ORD began
to list on its Web site human subject protection information available
through OHRP and other organizations and distributed a summary of lessons
learned from institutions that had been affected by recent sanctions by
regulatory agencies. ORD officials told us they expect to complete a draft
of a revised policy manual for VA research by September 2000.

ORCA officials have also implemented initiatives. For example, it began
bimonthly teleconference calls in February 2000 with IRB and research
officials at medical centers to share information and obtain input on human
protection issues. In March 2000, ORCA issued its first newsletter to local
research officials. This educational newsletter, planned as a twice a month
series, will address informed consent and human subject protection issues.
In April 2000, ORCA convened a group of VA research staff and outside
experts in human subject protections to identify training courses developed
elsewhere that VA could use. The group also plans to develop guidance and
strategies for VA to use to train IRB staff, members, and investigators.
Beginning in May 2000, ORCA sent the first of three notices to local
research programs alerting them of current human subject protection
concerns. In June 2000, it began issuing a monthly set of news clippings on
human subject protection issues.

In 1999, VA's National Center for Ethics sponsored a conference on ethics in
research and issued related reports including a discussion of principles for
researchers' consideration on the principles guiding the ethical conduct of
research involving participants with impaired capacity to consent. VA is
participating in national efforts to develop policies and procedures for
protecting these participants.15

Obtaining Information About Local Research Programs

VA headquarters officials have acknowledged that they lacked key information
about research programs at medical centers. To obtain more accurate and
complete information, they have taken several steps. Examples follow.

ï¿½ In October 1998, VA research officials began to develop a new computerized
data system to improve the comprehensiveness and accuracy of data about
studies involving human subjects at VA medical centers. As of June 2000,
development was still under way.

ï¿½ In April 1999, VA asked its medical centers whether they operated their
own IRB or relied on the IRB of an affiliated university. VA also asked
whether assurances with OHRP were involved. ORCA finished verifying this
information in July 2000.

ï¿½ In October 1999, ORD sent a questionnaire to the director of each Veterans
Integrated Service Network to assess the adequacy of staffing and support
for human subject protections at the medical centers in each network. A lack
of adequate resources was one of the three most common problems identified.
Sixteen of the 22 networks reported inadequate IRB support, including staff,
space, and equipment. Fourteen networks identified education as a priority
issue and cited the need for educational opportunities and guidance
documents. In May 2000, headquarters sent information to the networks on
educational opportunities and made suggestions for the level of
administrative staffing of IRBs.

ï¿½ By February 2000, VA had accepted an assurance from each medical center
conducting human research that it would comply with regulations for the
protection of human subjects.

ï¿½ In April 2000, VA's Chief Financial Officer reported that VA would
implement a system to allow for the explicit accounting of funds from the
medical care appropriation that are used by medical centers to support the
indirect costs of research.

These steps are necessary to obtain key information about human subject
research programs at medical centers. This information will allow
headquarters officials to determine the additional steps that may be needed
locally or systemwide to ensure compliance with regulations and the
protection of human subjects.

Implementing Organizational Changes to Enhance Monitoring and Oversight

VA is implementing two organizational changes to enhance its monitoring and
oversight of human research programs. The Under Secretary for Health
announced these changes in April 1999, but as of August 2000, they had not
been fully implemented. They are designed to allow routine onsite monitoring
of research programs, thereby helping medical centers identify weaknesses
and develop strategies to improve compliance with regulations and the
protection of human subjects. Although promising in concept, it is too soon
to determine whether the initiatives described below will fulfill their
objectives.

ï¿½ In April 1999, VA announced the creation of ORCA. VA did not begin
staffing this office until it appointed the chief officer in December 1999.
VA plans that ORCA will have eight headquarters staff by September 30, 2000,
and four regional offices with four staff each by December 31, 2000. As of
July 2000, VA had not completed its staffing of the headquarters component
and had not filled any regional office positions. Although ORCA's specific
plans for monitoring medical center research activities were still under
development in summer 2000, officials told us that they planned to conduct a
site visit on a rotating basis to each medical center conducting human
research. As of July 2000, ORCA officials told us they had not developed a
specific schedule for conducting these visits, but they expect to do so when
the regional offices are staffed. ORCA's headquarters has a budget of
$600,000 for fiscal year 2000 and $1.5 million for fiscal year 2001. The
regional offices have a budget of $1.9 million for fiscal year 2000 and $2.3
million for fiscal year 2001.

ï¿½ In August 2000, VA awarded a $5.8 million, 5-year contract for external
accreditation of its IRBs. This contract requires the contractor to conduct
a site visit every 3 years to each medical center conducting human research.
The contractor is expected to review IRB performance and to assess its
compliance with regulations. VA officials told us that VA expects that the
university-run IRBs it uses will grant access to the accreditation team. VA
is the first research organization to have an external accreditation of its
human research programs.

VA has not ensured that its medical centers have fully implemented required
procedures for the protection of human subjects. Primary responsibility for
implementation of these protections lies with local institutions?medical
centers and their IRBs. Although we cannot generalize from our sample to the
universe of VA research institutions, we found sufficient evidence of
noncompliance with applicable federal regulations to be concerned. We also
found that incomplete access to information about adverse events experienced
by research participants made it difficult for IRBs to fulfill their
mandate. We found widespread weaknesses in the management of human subject
protections that VA had not identified because of its low level of
monitoring. VA's past failure to ensure that its research facilities had the
resources, including staff, training, and guidance, needed to accomplish
their obligations suggests that headquarters has not given attention or
sufficient priority to the protection of human subjects.

Despite a 5-year record of problems at the West Los Angeles VA Medical
Center, VA did not begin to implement systemwide improvements until OHRP
took regulatory action against the medical center. VA's initial actions were
primarily crisis-driven and site-specific. Generally, appropriate corrective
actions have now been implemented at each of the three medical centers we
visited that were affected by regulatory sanctions. However, VA's progress
on systemwide improvements to its human subject protection system has been
slow. VA only recently began to obtain the information it needs--such as
identifying which medical centers use their own IRBs and which rely on
university-run IRBs?to plan necessary systemwide improvements.

Some facilities we visited and projects we reviewed appeared to have
reasonably strong protections for the rights and welfare of participants.
VA's recent efforts to improve its human subject protections systemwide and
its commitment to developing an effective oversight and monitoring system
are important steps toward ensuring that all VA facilities meet
requirements, but it is too soon to determine how well these initiatives
will fulfill their objectives.

VA has a long history of important contributions to medical research, and it
could set important precedents in improving human research protections. For
example, VA is the first federal agency to take action to externally
accredit its IRBs. Whether VA medical centers establish their own IRBs or
work with university-run IRBs, VA needs to ensure that the IRBs have
adequate resources, and VA must exercise its oversight authority if it is to
know what guidance, preventive efforts, or corrective actions are needed.

To strengthen VA's protections for human subjects, we recommend that the
Acting Secretary of Veterans Affairs direct the Under Secretary for Health
to take immediate steps to ensure that VA medical centers, their
IRBs--whether operated by VA or not--and VA investigators comply with all
applicable regulations for the protection of human subjects by

ï¿½ providing research staff with current, comprehensive, and clear guidance
regarding protections for the rights and welfare of human research subjects;

ï¿½ providing periodic training to investigators, IRB members, and IRB staff
about research ethics and standards for protecting human subjects;

ï¿½ developing a mechanism for handling adverse event reports to ensure that
IRBs have the information they need to safeguard the rights and welfare of
human research participants;

ï¿½ expediting development of information needed to monitor local protection
systems, investigators, and studies and to ensure that oversight activities
are implemented; and

ï¿½ determining the funding levels needed to support human subject protection
activities at medical centers and ensuring an appropriate allocation of
funds to support these activities.

In written comments (see app. II) on a draft of this report, VA agreed with
our findings and recommendations. VA said that initiatives it has already
planned and implemented will provide a foundation for a national prototype
in effective human subject protections.

Although VA agreed that its implementation of a systematic approach to human
subject protections has been slow to develop, it provided clarification
regarding statements in the draft report that VA had not focused attention
on systemwide weaknesses until after the March 1999 regulatory action at the
West Los Angeles VA Medical Center. VA stated that planning for the
establishment of regional offices for risk management and research
compliance had begun almost 1 year earlier. We have modified the report
accordingly.

In concurring with our recommendations to provide research staff with
current, comprehensive, and clear guidance and training about human subject
protections, VA identified initiatives planned or under way to improve its
guidance, disseminate the guidance, and train research staff in its use.
These initiatives represent promising efforts. Whether VA's plans for
guidance and training are effective will depend upon implementation details.
VA must ensure that its research staff have access to and receive current
guidance and training to enable them to meet their obligations to protect
the rights and welfare of human research subjects.

VA agreed with our recommendation to improve adverse event reporting and
said it has expanded the distribution of reports from its data monitoring
boards to include all appropriate IRBs. VA has also indicated its intention
to participate in governmentwide efforts to address this matter. These are
important first steps in ensuring that IRBs have the information they need
to safeguard the rights and welfare of human subjects. However, because the
VA monitoring boards analyze only those adverse events reported in VA's
multicenter Cooperative Studies program, further efforts to address reports
of adverse events from other studies are necessary.

VA also concurred with our recommendation to improve monitoring and
oversight of human subject protection activities and identified several
activities it has planned or implemented, such as external accreditation of
IRBs and establishment of performance measures related to human subject
protections for medical center research officials. Oversight and monitoring
are essential if VA is to know whether the procedures at its medical centers
and affiliated universities comply with human subject protection
regulations. Whether the actions VA plans to take in this area will be
sufficient depends on how effectively they are implemented.

Finally, VA concurred with our recommendation to determine the funding
levels needed to support human subject protection activities at medical
centers and then ensure an appropriate allocation of funds to support these
activities. VA's response notes that it has begun to account for medical
center expenditures associated with research support--an important first
step toward determining necessary funding levels. However, VA did not
discuss how it would ensure that funds are appropriately allocated to human
subject protection activities. As we noted, organizational tensions within
VA have created a situation in which there is no clear focus of
responsibility for funding such activities at medical centers. Until this is
addressed, we are concerned that VA cannot ensure that human subject
protections will be appropriately funded.

VA officials also provided technical comments, which we incorporated where
appropriate.

We are sending this report to the Honorable Hershel W. Gober, Acting
Secretary of Veterans Affairs, appropriate congressional committees, and
other interested parties. We will also make copies available to others upon
request.

Please contact me at (202) 512-7101 if you or your staff have any questions.
An additional GAO contact and the names of other staff who made major
contributions to this report are listed in app. III.
Cynthia A. Bascetta
Associate Director, Veterans' Affairs and
Military Health Care Issues

Scope and Methodology

Our objectives were to (1) assess the Department of Veterans Affairs' (VA)
implementation of human subject protections, (2) identify whether weaknesses
exist in VA's system for protecting human subjects, and (3) assess VA's
actions to improve human subject protections at those sites affected by
sanctions imposed by regulatory agencies and throughout VA's health care
system.

To achieve these objectives, we reviewed VA, Food and Drug Administration
(FDA), and Department of Health and Human Services (HHS) regulations and VA
policies for the protection of human subjects; interviewed VA research
officials; visited selected VA medical centers to assess local
implementation of these standards; and visited VA medical centers affected
by research restrictions. We also interviewed officials from the Office for
Human Research Protections (OHRP) and reviewed HHS guidance. We reviewed
records of congressional hearings; reports on human subject protections,
including those issued by the HHS Office of Inspector General, the Institute
of Medicine, and the National Bioethics Advisory Commission; and the
literature on the history of human subject protections.

To assess VA's implementation of human subject protections, we conducted
site visits at eight VA medical centers: Atlanta, Ga.; Baltimore, Md.;
Cleveland, Ohio; Dallas, Tex.; Louisville, Ky.; Providence, R.I.; Seattle,
Wash.; and Washington, D.C. We selected sites to reflect major differences
in VA research programs (see table 1). First, we selected medical centers
that differed in the number of studies they conduct with human subjects.
Second, we selected medical centers that differed in the institutions
responsible for operating the committee tasked with reviewing each study to
assess its protections for human subjects?the institutional review board
(IRB). Third, we selected facilities that differed in the assurance
arrangements they had with OHRP. Some institutions had filed a legally
binding commitment to comply with federal regulations called a multiple
project assurance with OHRP; other institutions had not. Our results from
these eight medical centers cannot be generalized to other sites.

                                       Number of
                                       VA                                Number
                                                                         of
              Institution    Multiple            Number   Number of      signed  Number
                                       human
 VA medical   responsible    project   research  of       signed consent consent of IRB
 center       for            assurance studies   studies  forms sought   forms   minutes
              operating      filed with          in our   from           sought  in our
              IRB            OHRP      reported  sample   investigators  in      sample
                                       in fiscal
                                       year                              medical
                                       1998a                             records
 Atlanta      Universityb    Yes       217       20       59             30      6
 Baltimore    Universityc    Yes       356       17       87             15d     5

 Cleveland    VA and         Yes       67        15       59             20      4
              Affiliatese
 Dallas       VA             Yes       235       20       92             32      5
 Louisville   VA             No        102       15       57             23      7
 Providence   VA             No        90        20       74             36      5
 Seattle      Universityf    Yes       354       22       72             29      5
 Washington,
 D.C.         VA             Yes       320       17       40             17      5

aThese data, obtained from VA's Research and Development Information System,
were the most recent data regarding the number of studies involving human
subjects available at the time we selected our sites.

bThe Atlanta VA Medical Center's IRB-of-record is that of Emory University.
At the time of our visit, Emory had two IRBs that together review about
2,000 projects each year.

cThe Baltimore VA Medical Center's IRB-of-record is that of the University
of Maryland, Baltimore. Its two IRBs together review about 1,200 projects
each year.

dConsent forms in Baltimore were in electronic rather than paper medical
records.

eThe Cleveland VA Medical Center is one of four cosignatories to a multiple
project assurance; the others are Case Western Reserve University, the
University Hospitals of Cleveland, and the MetroHealth System. Each of these
four institutions has its own IRB. Representatives of each IRB serve on a
single IRB advisory committee designed to promote discussion of concerns and
continuing improvement of the IRB system.

fThe Seattle VA Medical Center's IRB-of-record is that of the University of
Washington, which also serves the Boise VA Medical Center in Idaho. The
University of Washington has three IRBs, two restricted to medical research
and one restricted to behavioral research; together they review about 3,000
projects per year. The University of Washington has agreements with several
other institutions for specific IRB reviews. For example, in recognition of
the specialized expertise required for cancer research, the IRBs of Swedish
Hospital or the Fred Hutchinson Cancer Research Center review all such
protocols, including VA cancer studies. We obtained IRB records for one
project from each of those sites.

At each site, we interviewed local research personnel, including the
associate chief of staff for research and development, the IRB chair, and
staff responsible for providing administrative support to the IRB. We
attended an IRB meeting at six sites (Atlanta, Cleveland, Dallas,
Providence, Seattle, and Washington, D.C). We also reviewed written
procedures describing how the IRB and institution implement human subject
protections and a sample of four to seven sets of IRB minutes from the last
2 years (December 1997 through October 1999) at each site.

We randomly selected a sample of 15 to 22 projects at each site for detailed
analysis. To ensure that our selection included research on potentially
vulnerable participants, we oversampled studies designed to provide
information about psychiatric conditions that can affect decision-making
capacity, such as dementia, schizophrenia, and depression. Up to one-fourth
of the studies we sampled at any one site were in this category. We examined
IRB records for each project in our sample (146 in all, including 27
psychiatric studies). For the subset of 138 studies that required written
consent, we reviewed the most recently approved consent form.16 To determine
whether subjects had signed appropriate consent forms indicating willingness
to participate in research and whether those forms were available as
required, we examined about 5 signed consent forms maintained in
investigators' files from each of 125 studies. We also tried to obtain about
two signed consent forms from each project in paper medical records. This
sample included 98 projects. Some medical records could not be made readily
available to us. For example, some medical records were at a different
location during our visit.

To assess corrective actions at VA medical centers in response to
restrictions on their human research programs, we conducted 2-day visits to
three facilities where human research was suspended?Chicago Westside, Ill.;
Denver, Co.; and West Los Angeles, Ca.17 Our site visit team included an
expert in human subject protections under contract to us. For each of these
sites, we examined the OHRP and FDA reports associated with the restriction
of human research, action plans for resolving identified problems, documents
regarding current human subject operations, and the status of the research
program and human subject protections at the time of our visits (February
2000 and March 2000). We discussed these matters with medical center
officials and officials from IRBs at affiliated universities when they were
involved. In addition, we reviewed documents and interviewed officials from
two other medical centers?Birmingham, Ala., and Richmond, Va. These
facilities were also affected when the IRBs of their affiliated universities
were cited for noncompliance with federal regulations. Both have now
established their own IRBs.

We conducted our work between June 1999 and August 2000 in accordance with
generally accepted government auditing standards.

Comments From the Department of Veterans Affairs

GAO Contact and Staff Acknowledgments

Bruce D. Layton, (202) 512-6837

Cheryl Brand, Kristen Joan Anderson, Jacquelyn Clinton, Patricia Jones, and
Janice Raynor also made key contributions to this report. In addition, Barry
Bedrick and Julian Klazkin provided advice on legal issues, and Deborah
Edwards provided advice on methodological issues.

Related GAO Products

Medical Records Privacy: Access Needed for Health Research, But Oversight of
Privacy Protections Is Limited (GAO/HEHS-99-55).

Medical Records Privacy: Uses and Oversight of Patient Information in
Research (GAO/T-HEHS-99-70).

Scientific Research: Continued Vigilance Critical to Protecting Human
Subjects (GAO/T-HEHS-96-102).

Scientific Research: Continued Vigilance Critical to Protecting Human
Subjects (GAO/HEHS-96-72).

(406171)

Table 1: Characteristics of Sampled VA Medical Centers 47

Figure 1: Human Research Project Review and Approval Process
at VA Medical Centers 11

Figure 2: Noncompliance With VA Regulations at Eight Sites 12

Figure 3: Elements of Informed Consent Required by VA
Regulations 14

Figure 4: VA's Regulatory Criteria for IRB Approval of Research
Projects 20
  

1. The Veterans' Benefits and Services Act of 1988 authorized VA to
establish nonprofit research corporations at its medical centers to broaden
VA's ability to accept and manage private and non-VA public funds to support
research programs. In 1998, there were 87 nonprofit research corporations
associated with VA medical centers with total revenues of almost $122
million.

2. The Office for Human Research Protections (OHRP) is in the Office of the
Assistant Secretary for Health. HHS established OHRP in June 2000 to assume
the human subject protection functions of the former Office for Protection
from Research Risks (OPRR), which was part of NIH. In this report, we refer
to both organizations as OHRP. Actions taken before June 18, 2000, were
taken by OPRR.

3. FDA does not have a comparable system of assurances.

4. National Commission for the Protection of Human Subjects of Biomedical
and Behavioral Research, The Belmont Report: Ethical Principles and
Guidelines for the Protection of Human Subjects of Research, Apr. 18, 1979.
The commission was established by the National Research Act of 1974. Federal
regulations for the protection of human subjects reflect the recommendations
of the Belmont Report.

5. VA and HHS regulations share a common core, but HHS regulations include
additional requirements for research involving fetuses, pregnant women,
human in vitro fertilization, prisoners, and children. VA research funded by
other federal agencies having human subjects regulations, such as the
Department of Defense, is also subject to oversight authority by those
agencies. In most cases, the applicable federal regulations are essentially
the same as those for VA.

6. Within VA, the IRB is a subcommittee of a medical center's research and
development committee and is often called the Subcommittee on Human Studies.

7. VA regulations specify two exceptions to the requirement that VA treat
research-related injuries: (a) the injury results from the subject's own
noncompliance with study procedures or (b) the research is done by VA
contractors (38 C.F.R. section 17.85).

8. VA regulations allow legally authorized representatives to provide
consent to participate in research for incapacitated subjects.

9. Two university-run IRBs we visited re-review studies they have classified
as posing a high degree of risk every 6 months.

10. See, for example, HHS Office of Inspector General, Institutional Review
Boards: A Time for Reform, OEI-01-97-00193 (Washington, D.C.: HHS, June
1998).

11. VA regulations, 38 C.F.R. section 16.103(b)(5).

12. The absence of documentation does not necessarily indicate an IRB's
failure to ensure appropriate additional safeguards. For example, one study
discussed at an IRB meeting we attended addressed mania, a psychiatric
condition that could interfere with subjects' capacity to provide informed
consent. Because the investigator's application to the IRB had not addressed
this issue, an IRB member had contacted the investigator and confirmed that
plans for the study included a series of steps to ensure that only competent
subjects would be recruited. These plans were reported to the IRB in detail
before it voted to approve the study, but nothing was recorded in the
minutes about these recruitment procedures.

13. In 1995, VA created 22 Veterans Integrated Service Networks, a new
management structure to coordinate the activities of and allocate funds to
VA hospitals, outpatient clinics, nursing homes, and other facilities in
each region.

14. The medical center established such a board in 1995, but it met only
briefly. Although an interim board met several times in 1999 after OHRP's
regulatory action, its minutes state that it was unable to function due to a
lack of staff.

15. See VHA National Center for Ethics, Challenges and Change: Reports From
the Veterans Health Administration (VHA) Bioethics Committee (Veterans
Health Administration, Office of the Under Secretary, Washington, D.C.:
1999). VA's policy manual addresses a narrow set of circumstances related to
questions of capacity to consent, namely, cases in which a person has been
found to be incompetent. VA plans to develop guidance regarding consent when
ability to consent is in question such as when decision-making capacity
fluctuates or is declining. The National Bioethics Advisory Commission
(NBAC) has also addressed these complicated issues in Research Involving
Persons With Mental Disorders That May Affect Decisionmaking Capacity (NBAC,
Rockville, Md.: December 1998). VA is participating with the National
Science and Technology Council to consider how to implement NBAC
recommendations.

16. Eight projects in our sample did not require written informed consent
because they were exempt from IRB review (five studies) or because informed
consent procedures were appropriately waived for all subjects (three
studies).

17. The West Los Angeles VA Medical Center is now part of the VA Greater Los
Angeles Healthcare System.
*** End of document. ***