Commentary
Protecting Confidentiality in Human
Research
T
he validity of clinical research relies on accurate and truthful data collected
without fear of disclosure of sensitive information. To that end, Congress authorized the Secretary of Health and Human Services to issue certificates giving
investigators protection as intended by the statute to refuse to disclose the identity
of research subjects, even when required by an order of the court. The Certificate of
Confidentiality is intended to guarantee that vulnerable research populations can
participate in research without fear of disclosure. Some court cases, however, have
raised concerns about whether the Certificate provides the promised protections
and whether the application procedure is overly cumbersome (1, 2). In addition,
the Patriot Act further clouds the security provided to research information presumably secured by the Certificate. At the same time, the National Institutes of
Health (NIH) and local institutional review boards have been encouraging, and
sometimes requiring, researchers to obtain a Certificate in order to carry out
research. These issues have been the focus of several recent symposia at national
conferences (3; unpublished 2012 paper of L. Dame; unpublished 2012 paper of T.
Zarcone; unpublished 2011 paper of L. Beskow).
Approximately 1,000 new Certificates (a mean of 1,016 Certificates [SD5103] from
2009 to 2011) are granted annually by NIH. Most Certificates are granted by the
National Institutes on Drug Abuse (24%), of Mental Health (22%), of Alcohol Abuse
and Alcoholism (10%), and of Child Health and Human Development (12%). Given
the increasingly important role played by Certificates, we briefly review five court
cases analyzing the legal status of the Certificate, discuss remaining areas of
uncertainty, and offer specific recommendations that would further secure and
extend the Certificate’s protections.
Background
In 1970, Public Health Service Act section 301(d), 42 U.S. Code 241(d), authorized
the Secretary of Health and Human Services to issue a Certificate of Confidentiality
to protect the identities of individuals participating in drug abuse research. At the
time, drug-addicted individuals generally refused to participate in research out of
concerns about civil and criminal liability. Public Health Service Act section 301
(PHS Act 301) thus authorized “persons engaged in research on the use and effects
of drugs to protect the privacy of individuals who are the subject of such research by
withholding from all persons not connected with the conduct of such research the
names or other identifying characteristics of such individuals” (italics ours). A critical
addition noted, “Persons so authorized to protect the privacy of such individuals
may not be compelled in any federal, state or local civil, criminal, administrative,
legislative or other proceedings to identify such individuals.” In 1974, the act was
modified to include “mental health, including research on the use and effect of
alcohol and other psychoactive drugs” (4) and further expanded in 1988 to cover all
health research (5). This inclusion appears prescient given more recent concerns
regarding genetic testing.
466
ajp.psychiatryonline.org
Am J Psychiatry 170:5, May 2013
COMMENTARY
Legal Cases
Over the 40-year history of the Certificate, several legal challenges have reached
the courts. The major cases, their outcomes, and legal precedence are briefly
described below.
People v. Newman (6). In 1972, a witness to a homicide thought she recognized the
perpetrator as a fellow research subject at a methadone maintenance facility. A
subpoena was obtained to compel the clinic to produce their patients’ pictures
(used by the clinic to ensure their patients’ identity) to assist in the perpetrator’s
identification. As the project held a Certificate, the program’s director refused to
comply with the subpoena. Although the trial court held the director in contempt
for refusing to comply, this ruling was reversed on appeal in the New York State
Court of Appeals. The U.S. Supreme Court declined to hear the case. People had
argued that a 1972 statute amending section 42 of the Federal Code would allow the
court to compel the investigator to release the photographs. However, the court
ruled that since the 1972 amendment did not specifically repeal PHS Act 301, the
protections provided by PHS Act 301 remained as long as a Certificate had been
obtained. Thus, the appellate ruling supported the original intent of Congress
affording protection of a research subject’s identity through the Certificate.
New York v. Still (7). The defendant was charged with possession of a controlled
substance (methadone). The defendant told the police that he was a patient and
research subject at a methadone clinic
and thereby legally in possession of
We believe the Certificate is generally
methadone. The clinic refused to
confirm this claim, citing the proassumed to offer far more protection
tection of the Certificate. The appelthan it does.
late court ruled that by volunteering
his attendance at the clinic, the defendant waived the rights provided by the Certificate. This case established that
confidentiality provided by the Certificate was not protected if the identity of the
research subject was voluntarily disclosed.
Murphy v. Philip Morris, Inc. (8). The plaintiff took action against Philip Morris, asserting
that he contracted lung cancer from secondhand tobacco smoke exposure. Philip
Morris sought to compel the University of Southern California (USC) to disclose
raw data from a Certificate-protected study supporting the relationship between
secondhand smoke and lung cancer. While the court granted access to data from
several study participants with identities redacted, it also issued a protective
order stating that the defendant could not disclose the data and was “expressly
prohibited” from identifying study participants. Furthermore, upon completion of
their defense, all data were to be returned to USC. The court thereby attempted to
ensure the protection of the identities of the research subjects, although not their
data. This case emphasizes a critical distinction in the Certificate: while the identity
of the research subject is protected, their research data are not. Thus, PHS Act 301
only protects “the names or other identifying characteristics of … individuals.”
State of North Carolina v. Bradley (9). This more recent case has produced significant
concerns regarding the Certificate’s protection (1). The defendant was charged with
statutory rape, and his attorney requested a prosecution witness’s study records
from Duke University Health System. The request was granted, albeit with the order
that study documents remain confidential unless used at trial or sentencing. The
presiding judge and appeals court subsequently ruled to protect, then allow, and
Am J Psychiatry 170:5, May 2013
ajp.psychiatryonline.org
467
COMMENTARY
ultimately protect access to the records. However, the judge and appeals court
based their decisions on the belief that the defense was unlikely to find exculpatory
evidence, not because of the protections provided by the Certificate. Thus, the
protections offered by the Certificate were not addressed. Nevertheless, the initial
ruling required the university to turn over the research subject’s identity to the
court. However, the identity was never released to the opposing counsel and thus
could be considered “nondisclosed.”
Connecticut Superior Court for Juvenile Matters (Case Name Redacted). An
investigator at Yale
University informed the Connecticut Department of Children and Families of
concerns regarding a mother’s ability to care for her children. The investigator also
reported that the children were participating in a research study. Attempts by the
department to obtain study information were upheld, as the voluntary disclosure
by the investigator precluded the use of the Certificate to protect against further
disclosure. Again, the courts found that voluntary disclosure of the research
subject’s identity, either by the subject or the investigator, negated the protections
offered by the Certificate.
As the cases above illustrate, areas of legal ambiguity surround the Certificate.
While the identities of the participants have been protected (unless revealed by
the investigator or the participant), the research data have not been protected.
Questions regarding the Certificate’s protections primarily concern future court
decisions that may or may not occur in response to challenges that could come to
pass. Nevertheless, most attorneys and courts over the past 40 years have been
responsive to the Certificate and have abided by its protections. On the other hand,
discussions with our colleagues in the scientific community led to our concern that
the Certificate offers fewer protections than is generally believed.
Areas of Uncertainty
Several areas of uncertainty about the Certificate’s protections remain. First,
concerns that sponsoring institutions may not support the Certificate’s protections
in the presence of a legal challenge may have some theoretical validity, although
to our knowledge all institutions, to date, have supported the Certificate. In requesting a Certificate, the sponsoring institution agrees to “support and defend the
authority of the Certificate against legal challenges.” However, an institution’s commitment to appealing court rulings that undermine the Certificate’s authority has
been neither specified nor tested.
Second, the optimal method to ensure the confidentiality of research participation while providing clinical care continues to evolve. Some programs maintain
separate research and clinical charts. However, this method may result in study
participants receiving experimental treatment, unbeknownst to the treating
clinician. An alternative approach requires the clinical chart to reflect research
participation. For example, the Veterans Health Administration (VHA) requires all
participation in clinical research that utilizes VHA resources or that may result in
psychological or physical harm to be documented in the clinical chart (VHA
Handbook 1907.01 and 1004.1). While providing clinicians with information
necessary to treat the patient, this process allows widespread access to Certificateprotected information. Unfortunately, protection from being compelled to reveal
participants’ identifiers only extends to “persons engaged in … research”; it may
not extend to medical personnel aware of the patient’s participation but not
actually conducting the research. Although the Office for Human Research
468
ajp.psychiatryonline.org
Am J Psychiatry 170:5, May 2013
COMMENTARY
Protections (http://www.hhs.gov/ohrp/policy/certconf.html) states that the protections offered by PHS Act 301 include “researchers and institutions,” whether
these protections extend to all clinical employees of the institution has not been
tested. In addition, if the Certificate is obtained through an academic institution
but a research subject’s participation is noted in a VHA chart, do the Certificate’s
protections extend to an affiliated institution? As the federal government’s eHealth
Initiative allows greater access to medical charts, the number of nonresearch staff
with access will exponentially rise. Thus, clinical or administrative hospital staff aware
of a research subject’s participation may inadvertently provide their identity.
Once a participant’s identity has been revealed, the Certificate protects neither the
person’s identity nor his or her research data. The commitment of a sponsoring
institution to resist subpoenas for nonresearch staff has not been tested.
A third area of uncertainty is related to the USA Patriot Act, which was signed
into law in 2001 and extended in 2011. Title II of the act, entitled “Enhanced Surveillance Procedures,” provides greater powers of surveillance to several government
agencies. Section 215 (the “library records” provision) allows government agencies
to gather “tangible things,” including “books, records, papers, documents, and
other items,” during investigations into international terrorism. Although the NIH
Certificate of Confidentiality Kiosk notes that the Patriot Act does not affect the
Certificate’s protections, this has not, to our knowledge, been tested. Section 215
also states that “No person shall disclose to any other person … that the Federal
Bureau of Investigation has sought or obtained tangible things under this section.”
Thus, investigators who reveal to research participants that their identity has been
compromised or who otherwise impede the investigation (even by challenging the
order) would be in violation of the act. The act, however, provides protection to
anyone who complies with the order, presumably protecting the investigator and
institution from liability resulting from disclosing information protected by the
Certificate.
Recommendations
Given these concerns, research information
entered on a clinical chart should be minimized. Disclosures of research-related
diagnoses and treatment should be avoided unless they are important for clinical
care (e.g., participation in a medication trial). Institutions should develop policies
to protect against inadvertent disclosures of research information from a clinical
chart.
Researchers and institutional review boards.
Research is needed to explore what study participants actually
understand about the Certificate and how this understanding does or does not
affect their decision to participate. Investigators should consider rephrasing the
recommended language of the NIH Kiosk to better convey the limitations described
here.
The consent process.
The NIH Kiosk should clearly note the Certificate’s limitations: 1) protection
from being compelled to reveal participants’ identities may only extend to those
engaged in the research, 2) research information entered on a clinical chart may
compromise the Certificate’s protections, and 3) a participant’s identity may
possibly be released if requested for national security. Although the Kiosk notes
that “voluntary disclosure of information by study participants themselves or any
disclosure that the study participant has consented to in writing, such as to
insurers, employers, or other third parties” is permitted, it is not specifically stated
NIH.
Am J Psychiatry 170:5, May 2013
ajp.psychiatryonline.org
469
COMMENTARY
that even disclosure of research participation to a family member could potentially
void the Certificate’s protections. A spouse suing for divorce, for example, could
request research-related data if he or she had prior knowledge of the research
subject’s participation. Whether a court would accept this claim is untested.
Considering the limitations of the Certificate, lawmakers should
expand its scope to include 1) the protection of research data in addition to the
participant’s identity and 2) the protection of nonresearch staff from being
compelled to identify a participant’s identity and data. These changes would
require an act of Congress, but they would mitigate all of the above concerns except
those posed by the Patriot Act.
Policymakers.
Conclusions
The protections provided by the Certificate are substantially greater than those
offered by the Health Insurance Portability and Accountability Act of 1996 (HIPAA).
Although the Certificate has generally held up to legal challenges, some court
rulings have exposed vulnerability. While the outcome of future legal challenges
cannot be predicted, and concerns remain regarding the best method to implement protections in the clinical research environment, the Certificate appears
to have served its original purpose. Nevertheless, we believe the Certificate is
generally assumed to offer far more protection than it does. Its inclusion in the
consent process may mislead research subjects, as well as the investigators and
parent institutions, into a false sense of security. However, with the appropriate
caveats and modifications, the Certificate can become an even stronger tool for
protecting the rights of research participants.
References
1. Beskow LM, Dame L, Costello EJ: Research ethics: certificates of confidentiality and compelled disclosure of
data. Science 2008; 322:1054–1055
2. Gunn PP, Joiner SD: Certificates should be strengthened. Science 2009; 323:1289–1290; author reply 1290
3. Taylor SF, Wisner KL: Negotiating the path to approval: finding solutions to common IRB issues. Neuropsychopharmacology 2010; 35:S48–S49
4. Comprehensive Alcohol Abuse and Alcoholism Prevention, Treatment, and Rehabilitation Amendments, Pub
L No 93–282, sec 122(b)
5. Health Omnibus Programs Extension, Pub L No 100–607, sec 163
6. People v Newman, 32 NY2d 379, 298 NE2d 651, 345 NYS2d 502, 1973
7. People v Still, 80 Misc 2d 831, 1975
8. Murphy v Philip Morris, Inc, CV99-7155-RAP(CDCal), 2001
9. State v Bradley, 179 NC App 551, 634 SE2d 258, 2006 NC App LEXIS 1968, 2006
BRYON ADINOFF, M.D.
ROBERT R. CONLEY, M.D.
STEPHAN F. TAYLOR, M.D.
LINDA L. CHEZEM, J.D.
From the Department of Psychiatry, University of Texas Southwestern Medical Center, Dallas; VA North Texas
Health Care System, Dallas VAMC; Eli Lilly and Company, Indianapolis; University of Maryland at Baltimore;
the Department of Psychiatry, University of Michigan, Ann Arbor; the College of Agriculture, Purdue University,
West Lafayette, Indiana; and the Indiana Alcohol Research Center, Indiana University, Indianapolis. Address
correspondence to Dr. Adinoff (bryon.adinoff@utsouthwestern.edu). Commentary accepted for publication
December 2012 (doi: 10.1176/appi.ajp.2012.12050595).
Dr. Adinoff has served as a consultant for Shook, Hardy & Bacon LLP (medical malpractice consultant, tobacco
companies). Dr. Conley is a full-time employee of and shareholder in Eli Lilly. Dr. Taylor receives research
support from St. Jude Medical and Neuronetics. Dr. Freedman has reviewed this commentary and found no
evidence of influence from these relationships. Ms. Chezem reports no financial relationships with commercial
interests.
470
View publication stats
ajp.psychiatryonline.org
Am J Psychiatry 170:5, May 2013