Trust, confidentiality, and the acceptability of sharing HIV-related patient data: lessons learned from a mixed methods study about Health Information Exchanges

-

Documents
14 pages
Obtenez un accès à la bibliothèque pour le consulter en ligne
En savoir plus

Description

Concerns about the confidentiality of personal health information have been identified as a potential obstacle to implementation of Health Information Exchanges (HIEs). Considering the stigma and confidentiality issues historically associated with human immunodeficiency virus (HIV) disease, we examine how trust—in technology, processes, and people—influenced the acceptability of data sharing among stakeholders prior to implementation of six HIEs intended to improve HIV care in parts of the United States. Our analyses identify the kinds of concerns expressed by stakeholders about electronic data sharing and focus on the factors that ultimately facilitated acceptability of the new exchanges. Methods We conducted 549 surveys with patients and 66 semi-structured interviews with providers and other stakeholders prior to implementation of the HIEs to assess concerns about confidentiality in the electronic sharing of patient data. The patient quantitative data were analyzed using SAS 9.2 to yield sample descriptive statistics. The analysis of the qualitative interviews with providers and other stakeholders followed an open-coding process, and convergent and divergent perspectives emerging from those data were examined within and across the HIEs. Results We found widespread acceptability for electronic sharing of HIV-related patient data through HIEs. This acceptability appeared to be driven by growing comfort with information technologies, confidence in the security protocols utilized to protect data, trust in the providers and institutions who use the technologies, belief in the benefits to the patients, and awareness that electronic exchange represents an enhancement of data sharing already taking place by other means. HIE acceptability depended both on preexisting trust among patients, providers, and institutions and on building consensus and trust in the HIEs as part of preparation for implementation. The process of HIE development also resulted in forging shared vision among institutions. Conclusions Patients and providers are willing to accept the electronic sharing of HIV patient data to improve care for a disease historically seen as highly stigmatized. Acceptability depends on the effort expended to understand and address potential concerns related to data sharing and confidentiality, and on the trust established among stakeholders in terms of the nature of the systems and how they will be used.

Sujets

Informations

Publié par
Publié le 01 janvier 2012
Nombre de visites sur la page 24
Langue English
Signaler un problème

Maiorana et al. Implementation Science 2012, 7:34
http://www.implementationscience.com/content/7/1/34
Implementation
Science
RESEARCH Open Access
Trust, confidentiality, and the acceptability of
sharing HIV-related patient data: lessons learned
from a mixed methods study about Health
Information Exchanges
*Andre Maiorana , Wayne T Steward, Kimberly A Koester, Charles Pearson, Starley B Shade,
Deepalika Chakravarty and Janet J Myers
Abstract
Background: Concerns about the confidentiality of personal health information have been identified as a potential
obstacle to implementation of Health Information Exchanges (HIEs). Considering the stigma and confidentiality
issues historically associated with human immunodeficiency virus (HIV) disease, we examine how trust—in
technology, processes, and people—influenced the acceptability of data sharing among stakeholders prior to
implementation of six HIEs intended to improve HIV care in parts of the United States. Our analyses identify the
kinds of concerns expressed by stakeholders about electronic data sharing and focus on the factors that ultimately
facilitated acceptability of the new exchanges.
Methods: We conducted 549 surveys with patients and 66 semi-structured interviews with providers and other
stakeholders prior to implementation of the HIEs to assess concerns about confidentiality in the electronic sharing
of patient data. The patient quantitative data were analyzed using SAS 9.2 to yield sample descriptive statistics. The
analysis of the qualitative interviews with providers and other stakeholders followed an open-coding process, and
convergent and divergent perspectives emerging from those data were examined within and across the HIEs.
Results: We found widespread acceptability for electronic sharing of HIV-related patient data through HIEs. This
acceptability appeared to be driven by growing comfort with information technologies, confidence in the security
protocols utilized to protect data, trust in the providers and institutions who use the technologies, belief in the
benefits to the patients, and awareness that electronic exchange represents an enhancement of data sharing
already taking place by other means. HIE acceptability depended both on preexisting trust among patients,
providers, and institutions and on building consensus and trust in the HIEs as part of preparation for
implementation. The process of HIE development also resulted in forging shared vision among institutions.
Conclusions: Patients and providers are willing to accept the electronic sharing of HIV patient data to improve care
for a disease historically seen as highly stigmatized. Acceptability depends on the effort expended to understand
and address potential concerns related to data sharing and confidentiality, and on the trust established among
stakeholders in terms of the nature of the systems and how they will be used.
Keywords: Trust, Confidentiality, Acceptability, Health information exchanges, HIV, Patient data-sharing
* Correspondence: andres.maiorana@ucsf.edu
Center for AIDS Prevention Studies, AIDS Policy Research Center, University of
California, 50 Beale St, suite 1300, San Francisco, CA 94105, USA
© 2012 Maiorana et al.; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative
Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and
reproduction in any medium, provided the original work is properly cited.Maiorana et al. Implementation Science 2012, 7:34 Page 2 of 14
http://www.implementationscience.com/content/7/1/34
Background confidence in the objective security measures in place to
Health information exchange (HIE) systems protectpersonaldata.Onthe other
hand,usersneedtoasHealth information technology (HIT) has been identified sign a value and expect a benefit from the exchange
sysas a critical tool for improving medical care and treatment tem. Users make determinations about the value of
while holding down costs [1-3]. HIT has the potential to electronic data sharing within a larger social and
psychoimprove the quality and efficiency of medical care through logical context that affects how they navigate mutual
intermore informed decision making, enhanced communica- actions and calculate benefits and potential risks of their
tion with patients, better tracking of clinical indicators and choices[27]. Building a ‘framework of trust’among the
difmedical records, and improved management of databases ferent entities participating in a HIE related to the privacy
about diseases and treatments [1,3-9]. Within the United and confidentiality ofdata sharing [28],and adherence toa
States (US), HIT has become an essential component of ‘common framework’ that includes basic policies and
stanefforts to reform the healthcare system, including support dards have been proposed to overcome challenges to HIE
within the American Recovery and Reinvestment Act of implementation [29].
2009 for the adoption of electronic medical records and Early research documented substantive concerns about
within the Health Information Technology for Economic the security of personal health information stored in
and Clinical Health Act to aid in the development of a na- electronic systems [30-34]. However, recent studies
sugtionwide HIT infrastructure[10]. gest a growing acceptance among healthcare providers
In order to facilitate information flow across the and patients of health technology [35,36], and greater
healthcare system, provider groups, hospitals, insurers, willingness to share information through HIEs [37,38].
and governments are taking steps to implement HIEs, Because such acceptance is enhanced when patients and
which are characterized by formal agreements and tech- providers are informed or involved in a system’s
developnologies that facilitate the electronic movement of ment [39-41], experts emphasize the importance of
conhealth-related information across organizations within sidering attitudinal issues, such as buy in and trust from
an area or community [11]. Despite their promise, there stakeholders, including patients, when developing new
are significant barriers to widespread adoption of health HIEs [42,43]. Different theoretical constructs, including
information technology and exchange. Recent research several variations of the Technology Acceptance Model
has shown a relatively low uptake of electronic records (TAM) [44], have been used to understand what is
nesystems in both private provider and hospital settings cessary for the successful implementation and
acceptabil[12,13]. Among the obstacles are financial and personnel ity of technological systems. Tung et al. propose to add
investments [14], development and adherence to com- trust as another dimension to the TAM model
influenmon industry-wide protocols to facilitate data transfers cing beliefs and attitudes, intention, and perceived
useamong systems [15,16], and concerns about the confi- fulness to use [45]. However, while trust in electronic
dentiality of personal health information [5,17-20]. systems’ privacy and security measures, if accompanied
by oversight and stronger accountability mechanisms,
could be a facilitator to HIE implementation and adop-Trust in HIEs and confidentiality of electronic data sharing
tion [42,46], there is limited research on the impact ofTrust, in general, implies certain vulnerability and the
betrust and how attitudes toward privacy and confidential-lief or expectation that actions or social relationships will
ity may influence the development and implementationcause no harm and/or will provide a benefit [21]. Not
surof HIEs and the integration of information technologiesprisingly, trust is a key factor in the delivery of healthcare,
with human immunodeficiency virus (HIV) care.as research has shown that high levels of provider/patient
trustisconducivetomoreeffectivehealthcare[22].
Consequently, the building of trust in healthcare settings has Purpose of this analysis
emerged as a central concern [23-25]. Trust in healthcare HIV care is an important but challenging sector for the
can be seen as a three-part relationship between patient rollout of information technologies. Given the
complex(truster), provider or organization (trustee), and the spe- ities of treating HIV disease, information exchange would
cific context of delivering healthcare. In that context, trust be expected to be useful for improving care outcomes and
also relates to the expectation that an entity will adhere to delivering multiple services. New information technologies
their fiduciaryobligations[26]. have already demonstrated success in HIV care settings
According to Heuwinkel and Deiters, trust factors and [47], including improved care coordination [48] and
attitudes related to acceptance of HIEs operate both at an enhanced provider satisfaction [49], as well as increased
objective and a subjective level and are influenced by the acceptability among patients [50]. However, the relatively
interplay of technological, sociological and psychological high
levelsofstigmatowardthediseasehavegreatlyheighissues. On the one hand, users need to trust in the techno- tened confidentiality concerns around HIV [51], resulting
logical elements of the exchange system and have in HIV-related patient data historically enjoying greaterMaiorana et al. Implementation Science 2012, 7:34 Page 3 of 14
http://www.implementationscience.com/content/7/1/34
legal protections than other kinds of medical information information and HIV medical care. In addition, the
[52]. patients were asked to agree or disagree with a series of
Considering the stigma and confidentiality factors his- statements indicating willingness to have health
informatorically associated with HIV disease, we examine in this tion electronically shared with various providers, payers,
article how trust—in technological systems, operational and agencies: their primary HIVcare provider; other
cliniprocedures, and people—influenced the acceptability of cians in the clinic of the primary HIVcare provider; other
data sharing among patients, providers, and other stake- non-clinical staff in the clinic of the primary HIVcare
proholders before implementation of six distinct HIEs in dif- vider; non-HIVspecialists(e.g.,cardiologists,ob/gyn);other
ferent regions around the US. Our analyses identify the healthcare providers (e.g., emergency or hospital
concerns expressed by stakeholders about the confidenti- personnel); pharmacists; HIVsupport service organizations
ality of electronically shared data and focus on the fac- (e.g., case managers); other non-HIV-specific support
sertors that ultimately facilitated acceptability of the HIEs. vice organizations (e.g., drug treatment programs, mental
health programs); private health insurers, government
health insurers; and local health departments.For example,Methods
two specific survey items were ‘I am willing to allow myThe Health Resources and Services Administration HIV/
personal health information to be shared with my primaryAIDS Bureau (HRSA/HAB) sponsored a four-year Special
HIVcare provider (e.g., physician, nurse practitioner,
phys-ProjectofNationalSignificance(SPNS)knownastheElecician assistant) using a secure electronic network,’ and ‘Itronic Networks of Care Initiative. It consisted of one
am willing to allow my personal health information to becross-site evaluation center at the University of California
San Francisco (UCSF) and six unique demonstration sites shared with my private health insurers using a secure
elecindependent from each other implementing HIEs tailored tronic network.’ Responses were recorded on a five-point
to the local HIV epidemic and the care systems in those scale ranging from strong disagreement (0) to strong
locations. These HIEs were designed to enhance HIVcare, agreement (4).
promote the flow of HIV-related health information across
clinical settings, community-based organizations, and pub- Qualitative interviews
lic health agencies, as well as to improve patients’access to UCSF investigators experienced in qualitative research
their health records. For that purpose, each HIE created conducted the interviews either face-to-face during site
differentkindsof electronic platforms tosharepatient data visits or over the phone. We conducted a total of 66
among the collaborating medical settings and outside semi-structured interviews with three types of key
inagencies. Table 1 presents the key characteristics of the six formant stakeholders: project staff and information
techHIEs. nology (IT) specialists associated with the demonstration
site project (n=22); staff from community-based
organiData collection zations and public health agencies collaborating in the
We present quantitative and qualitative data collected at design and implementation of the HIEs (n=12); and
each of the six sites prior to HIE implementation. Within intended users of each HIE (n=32), such as medical
prohealth services research, a mixed methods approach that viders and staff in clinical settings, as well as social
workintegrates qualitative and quantitative data can help to ers, case managers, and medical social workers at
examine the complexity of the healthcare environment, agencies external to the medical settings but linked to a
health-related issues, measure outcomes, and evaluate HIE. To ensure diversity of perspectives, we collaborated
interventions [53]. The data were collected between Au- with site staff to identify stakeholders to be included in
gust 2008 and April 2010 as part of the UCSF multi-site our sample. The sample size at each site varied according
evaluation. The study was reviewed and approved by the to the project and the number of stakeholders involved
Institutional Review Board (IRB) at UCSF and by the in each of the HIEs.
corresponding IRBs at the six implementing sites. Using different semi-structured guides parallel in
content for each of the three types of key informant
stakeQuantitative surveys holders mentioned above, the interviews focused on
We administered a quantitative survey to HIV positive understanding the development of the HIE systems, the
patients at each site to obtain a description of the patient planning process and preparatory work, and the
populationand measuretheirwillingness to shareinforma- expected benefits, as well as the technological,
attitution using HIE. A convenience sample of patients (n=549) dinal, and structural barriers and facilitators to
acceptwas recruited by site staff at HIV clinical care settings ability of data sharing, including issues related to trust
involved with each HIE. These patients were administered and confidentiality. We specifically asked all stakeholders
an Audio Computer-Assisted Self-Interview (ACASI) sur- what confidentiality-related issues they expected or had
vey, which included questions about demographic had to address, including their own concerns as well asMaiorana et al. Implementation Science 2012, 7:34 Page 4 of 14
http://www.implementationscience.com/content/7/1/34
Table 1 Description of the six Health Information Exchanges part of the Electronic Networks of Care Initiative
Site Collaborating Institutions Users Goals and Scope of the HIE Description of Information Exchanged
Bronx-Lebanon - Hospital based HIV - Clinical providers Clinic-based system to improve ? Bi-directional limited in real-time information exchange
Hospital Center specialty clinic coordination of care and health
? Medical providers have access to reminders on clinical
outcomes
- Support services - Support services tasks/priorities and patient clinical indicators are
(case managers, exchanged with case management and medical support
medical social service agencies.
workers, nurses)
City of Paterson* - Primary care clinics - Clinical providers Localized area newtwork of care ? Patient information is exchanged through a Web-based
to exchange information and quality system in real time among administrators, medical providers
- Support services - Clinic staff
tools to enhance efficiency, continuity, and staff in primary care clinics, support service agencies, and
and quality of HIV care HIV testing sites for quality assurance and to improve quality
-Testing sites -Administrators
of care and patient satisfaction.
Duke University* - Hospital based - Clinical providers Localized area newtwork of care to ? Information uploaded and shared through a regional server
primary care enhance efficiency, continuity, quality
? Patient clinical indicators are exchanged among a medical
clinic and delivery of HIV care among partner
setting and case management support service agencies.
agencies within a RHIO
- Support services - Support services (case
managers and social workers
Administrators
LSU Health Care - Emergency departments, - Clinical providers Public health exchange to improve ? Real-time, bidirectional information exchange between
Services Division - Outpatient and inpatient HIV case reporting, identify and link a public health jurisdiction and a health care delivery system
clinics in 7 public hospitals to care HIV+individuals out of care
or lost to care
- Office of Public Health - Surveillance and ? Alerts are exchanged to identify out of care HIV infected
public health personnel individuals who seek treatment for other conditions and link
them to HIV care.
- Louisiana Public -Administrators
Health Institute
New York- Presbyterian - HIV care providers - Patients Localized area newtwork of ? Web-based system
Hospital* - Support services - Clinical providers care to improve coordination and
? Provider and patient portals allow patients access to their
- Health Insurer - Support services increase quality of care
own information, and allow medical providers, administrators,
Administrators
and social workers, case managers, and outreach workers
from support service agencies to exchange patient information.
St. Mary Medical Center - Primary care clinic - Clinical providers Clinic-based to improve patient ? EMR’s interoperable, bidirectional exchange
Foundation quality and efficiency of care
- Laboratory - Clinic staff ? Patient laboratory requisitions and results as well as medication
prescriptions and refills are exchanged among clinic,
- Pharmacy - Pharmacies
laboratory, and pharmacy staff to avoid duplication of services
and facilitate communication
- Laboratories
* Involved in fiduciary relationship among partners.Maiorana et al. Implementation Science 2012, 7:34 Page 5 of 14
http://www.implementationscience.com/content/7/1/34
those of the patients. Probes were used to prompt infor- Quantitative data from patients
mation that was not spontaneously offered by the partici- As reflected in Table 2, the 549 patients in our sample were
pants and further enquire into topics of interest. For HIE diverse in terms of race, ethnicity, gender, sexual
orientasystem users, we specifically conducted interviews after tion, education, and risk factors for HIV. Figure 1 displays
they had been trained on how to use the systems, but be- the mean levels and the corresponding 95 % confidence
fore those systems were implemented. All interviews limits of patient willingness to share medical information
were audio-recorded and transcribed. electronically with different types of providers, payers, and
institutions. Patients were most comfortable sharing health
information with their primary care provider, followed byData analyses
other clinicians, other healthcare providers, HIV supportWe initially analyzed quantitative and qualitative data
organizations, andgovernmenthealth insurers.Atthe otherseparately. Subsequently, the two datasets were
compared, with the quantitative findings from the patient
survey helping to inform and better frame qualitative
Table 2 Descriptive characteristics of the patients in the
findings from providers and other stakeholders. quantitative sample (N=549)
Characteristics Mean 95 % Confidence
IntervalQuantitative data
We cleaned the data as needed and generated descriptive Age (yrs.) 45.3 44.5, 46.1
statistics for the sample. Following this, the mean levels n%
of agreement and the corresponding 95 % confidence
Gender
limits for each of the questions about willingness to
Male 347 63.2
share medical information via HIE were calculated. All
Female 197 35.9analyses were conducted using SAS V9.2.
Transgender 5 0.9
Race / EthnicityQualitative data
Caucasian 113 20.6All interviews transcripts were entered into Atlas ti©, a
software program designed to organize qualitative data African-American 285 51.9
and to facilitate analysis. Data analysis procedures fol- Hispanic / Latino 106 19.3
lowed an open-coding process developed by Strauss and
Mixed 29 5.3
Corbin [54]. During the initial phase of analysis, three
Other 16 2.9
analysts individually read a subset of interviews and
Educationdeveloped preliminary codes based on domains from the
Less than high school 179 32.6interview guides and emerging concepts and categories.
The analysts refined those preliminary codes and dis- High school or G.E.D 240 43.7
crepancies in coding were solved during team meetings. Greater than high school 130 23.7
The final version of the codebook consisted of 16 coding
Annual Income
topics, which the analysts then applied across all the
No income 127 23.1
interviews. Each interview was coded by a primary
anaLess than $5000 121 22.0lyst and verified by a secondary analyst. Coded data were
Between $5000 and $10000 165 30.1summarized within each of the six sites. Convergent and
divergent perspectives were then examined within and Between $10000 and $20000 87 15.9
across sites. Analyses for this article included all data Greater than $20000 49 8.9
associated with the codes ‘confidentiality,’‘process of
colSexual Orientation
laboration,’‘challenges,’ and ‘expected benefits.’ The rest
Heterosexual 320 58.1
of the codes, while relevant for the overall evaluation of
Homosexual 156 28.4the HIEs, did not apply to trust or confidentiality issues,
Other 73 13.3and are thus not considered here.
Risk Factors for HIV
MSM 186 33.9Results
We first present quantitative data reflecting on the IDU 63 11.5
patients’ willingness to share personal health information Heterosexual contact 192 35.0
electronically. We then present findings from the
qualitaOther 108 19.7
tive interviews with providers and other stakeholders on
Homeless 120 21.9
their viewsand acceptabilityofdata sharing throughHIEs.Maiorana et al. Implementation Science 2012, 7:34 Page 6 of 14
http://www.implementationscience.com/content/7/1/34
Figure 1 Patients’ level of willingness to share medical information electronically with various recipients. Note: 0 indicates low willingness
and 4 indicates high willingness to share information electronically.
end of the spectrum, patients were less willing to share information to be shared electronically. Those perceptions
health information with private health insurers and other correspond with the patients’ willingness to share data
non-clinical staff. The mean levels of willingness in every shown in the quantitative data. Stakeholders identified
facategory fell above the mid-point of the measurement scale miliarity with the use of electronic technology, trust in the
(which was defined as the point where patients neither institutions and in the staff providing services, and the
agreed nor disagreed with sharing information), reflecting expected benefit of HIE as facilitators of the acceptability
an overall comfort with electronically sharing information. ofdata sharing amongpatients.
Stakeholders identified a variety of factors that they
Qualitative data from project staff, IT specialists, and thought contributed to overall patient comfort with data
providers sharing. One of the primary factors was the growing
acThese findings are organized to present stakeholders’s ceptance of health-related IT, a result of the increasing use
of electronic records systems in clinical settings. Patientsperspectives at four different levels: their views and
have become used to seeing their information reviewed inexpectations about patients’ acceptability of data-sharing;
orentered intoa computer whilereceiving services:their own views regarding of patient data
sharing; their views and experiences that reflect upon
ac‘And again regarding electronic mistrust issues.... we’veceptability at an institutional level; and trust as an
overkind of progressed over the past five or six years toarching factor influencing the stakeholders’ views on
using our electronic, internal electronic medicaldata sharing. Table 3 shows the main factors influencing
acceptability included in the qualitative findings. record system a lot more and so patients are very used
to seeing people typing... as soon as you show up
things are going into computers and... allPatients see HIEs as a change in the mechanism but not
prescriptions are electronic and we usually are usingin the practice of exchanging data
the computers while we’re talking to people so myIn general, stakeholders anticipated no major concerns
guess is that some of that’s not as much of an issue asfrom patients about sharing data. The consensus was that
it has been in the past.’—Medical Directormost patients would be willing to allow their healthMaiorana et al. Implementation Science 2012, 7:34 Page 7 of 14
http://www.implementationscience.com/content/7/1/34
Table 3 Main factors influencing stakeholders’ acceptability of electronic data sharing
Different Levels of Stakeholder Views Factors Influencing Acceptability of Data Sharing
Stakeholder views on ? patient familiarity with electronic technology
patient acceptability
? patient trust in institutions and staff
? expected benefits for patients
Stakeholder views on ? trust in the security systems to protect data
stakeholder acceptability
? greater understanding of extent of information sharing
? greater of parameters of access by differing individuals
Stakeholder views on confidentiality and ? establishment of legal foundation for data sharing and contractual agreements
acceptability at the institutional level
? trust in system security
? trust in HIE agencies and staff, partially dependent on agency precedents for sharing
electronic data sharing
Importance of trust in ? trust between patients and providers
establishing acceptability
? trust different stakeholders and between agencies
? trust in system security
Stakeholders also reported that, to patients, HIEs rep- opposition to electronic data sharing through statements
resent a change only in the mechanism but not in the such as ‘well, I don’t want you to know my business,’ or
practice of exchanging data. Stakeholders spoke of how ‘how dare you be able to do that,’ the same case manager
patients knew that their information was already shared expected clients to ask,‘who else has access to this
inforby fax, phone, or paper, and that this awareness contrib- mation,’ and to help her clients ‘...understand that we’re
uted to the acceptability of electronic exchange of infor- on this system together to assure quality of care more
mation. Similarly, familiarity among patients with the than anything else.’
concept of having to sign a release of information in Stakeholders reported that sites had conducted
formaorder for their information to be shared with external tive work pre-development of HIEs and consulted patients
medical providers or support agencies also facilitated ac- regarding their views on electronic data sharing. In one
ceptability of electronic data sharing: site, stakeholders stated that needs assessments had shown
thatpatients thoughtinformationwasalreadybeingshared
‘I guess because ofthe way I’veshareditwiththe with other agencies in an effort to provide better care.
clients...,tothemit’s just another way...another Other sites, also taking into account the opinions of
technology....we alreadyshareitanyway....it’s thesame patients, took a number of steps to control access to
electhing as me calling the doctor’sofficesayingcanyou tronic health information, with greater degrees of access
give ittome, as mepulling itupinthe system. I think generally being reserved for personnel most directly
they view itlikewe alreadyknow, ...we’regoingto involved in clinical care. For example, at one site, a
deciknowanyway. ... it’s hard formetoservice youas a sion was made not to give clerical staff access to detailed
client whenI don’t knowwhat’s going onwith you clinical information in electronic reminders after a patient
medically. Sothathelpsme togiveyou what you need advisory board raised concerns about privacy. In another
and ithelps itgoover a whole lot faster.Sothey don’t site, clerical staff registering patients were given alerts that
lookat isas a threat of ‘myinformation’sgoingtobe only told them that patients needed to talk to a medical
given out’ where ‘it’s mycasemanager and it’smy provider about further testing, without specifying what
doctor so it’scool”—Administrator and Case Manager. kind of testing was needed. And at a third site,
stakeholders reported that patients only became comfortable
Stakeholders emphasized that patients’ acceptance of with the idea of information being exchanged among
partHIE could be facilitated by being transparent about a ner agencies once they knew that unique identifiers and
system’s purpose, the types of information to be notnames wouldbeused inthe system.
exchanged, the expected benefits, and the individuals Nevertheless, a few of the users, based on their direct
and/or agencies that would have access. One case man- experience working with patients, noted that a small
ager anticipated that, once the local HIE went live, a line number of ‘cautious’ patients might either object to or
would be included on the release of information forms would need additional reassurance before they agreed to
so that clients would know that health from data-sharing. According to them, such patients were
the medical setting would be electronically available to likely to include older African Americans, substance
support services agencies. Rather than encountering users, undocumented Latinos and other groups that mayMaiorana et al. Implementation Science 2012, 7:34 Page 8 of 14
http://www.implementationscience.com/content/7/1/34
be more suspicious of the healthcare system. In addition, confidentiality had been addressed, stakeholders needed
objections were anticipated from patients who are very to have a clear understanding of a HIE’s purpose and
secretive about their HIV status or were categorized as intended uses, and the security measures to protect
pa‘hard to work with.’ Related to this issue, a clinic coord- tient data. For example, our interviewees explained that
inator stated, ‘There are a few patients who are just very stakeholders wanted to know about the access rights to a
cautious and concerned about their health information HIE, such as who would be able to log in and how, and
and...[their] needs are addressed on a one-to-one basis ... the extent of the information to be exchanged. Such
but as a rule it’s not something that we deal with on a knowledge helped assuage stakeholders’ concerns and
day to day basis.’ contributed to their personal buy-in to the system. That
knowledge also equipped them with information they
could use to educate patients about the HIE. Further-Project staff and users’ trust in the systems and the
more, as stated by project staff, users’ buy-in was import-security of the systems
ant because the trust of the users in the system willStakeholders were willing, and in many cases
enthusiasreflect in how they explain it to the patients, thus influ-tic, to share patient data electronically because they
encing the patients’ trust in the HIE, ‘because part ofexpected that it would make their work more efficient
what we do is we help case managers really feel like thisand be of benefit to patients or clients. Their acceptance
is doing what’s best and I think that translates to theof data sharing was facilitated by trusting the systems
patients; that trust translates.’and the security measures in place to protect the data, as
Paralleling expectations about patient attitudes, stake-well as by learning about the extent of the information
holders’ personal acceptance of the HIEs was also influ-to be shared and who would have access to the
informaenced by the perception that they represented only onetion exchanged. Their perspectives were influenced not
step beyond ‘business as usual,’ and sharing informationonly by knowledge about technological protections of
on paper or by phone. For example, one medical pro-data (e.g., encryption), but also by attitudes about the
apvider noted that electronic exchange simply provides apropriate extent of information to be shared and the
different mode of communication but that the funda-types of individuals who would have access
mentals of sharing clinical information remained theFor the most part, their potential concerns about
consame. Stakeholders pointed out that, at all agencies, pro-fidentiality had been resolved during the HIE design,
decedures already existed to ensure patient privacy. Manyvelopment, and training prior to our interviews.
partner agencies within the HIEs already were reportingStakeholders reported that their approaches to
addreseither identifiable or de-identified patient information tosing those concerns depended on the scope of the
prolocal medical consortia, to state or local health depart-ject, collaborative processes among HIE partners, and
ments for surveillance purposes, or to funding agenciesthe needs assessments the sites had conducted when
as a condition of grants that supported clinical opera-designing their systems. For example, one site conducted
tions. One medical provider pointed out that the pro-a comprehensive legal and ethical review to determine
blems that may lead to unintended lapses inwhat information could be exchanged and with whom. A
confidentiality (e.g., human error) could occur both withcollaborator part of that review stated:
electronic or non-electronic exchange of healthcare
information:‘And I can summarize everything that I’ve been doing
as falling into the category of trying to maximize the
‘...I’m sure there are ways that somebody could getinformation sharing that takes place and maximizing
into the computer system and get confidentialthe benefits to everyone...while still observing any
medical records. In faxing prescriptions thingsapplicable privacy restrictions, which would come
could probably be accidentally faxed to the wrongfrom state law or federal law or any other applicable
fax number....’ —Medical Providerstandard....I’m trying to help them get the most out
of this, you know, without legally violating anyone’s
privacy.’
Acceptability of data sharing at an institutional level
That review helped to alleviate stakeholder concerns Acceptability of data involved overcoming
bararound confidentiality by establishing clear legal bound- riers about privacy issues and potential liability for the
aries for the project’s scope and types of information to institutions connected with the HIEs. Institutions’
acbe exchanged. The formative work with patients pre-de- ceptance was facilitated by reviews regarding the legal
velopment of HIEs mentioned earlier also helped allevi- aspects of data sharing for the institutions, as well as by
ate stakeholders’ concerns regarding electronic data trust in the security of the systems, trust in the staff and/
sharing. In order to feel like concerns about or agencies involved in the HIEs, and whether there wasMaiorana et al. Implementation Science 2012, 7:34 Page 9 of 14
http://www.implementationscience.com/content/7/1/34
a precedent within the institutions sharing data ‘...we had togothrougha privacy analysiswith [nameof
electronically. company]...;and they havetheir ownlawyersandmost
A number of stakeholders, particularly project and IT everything that we needed wasalreadywritten up forus
staff, reported having spent considerable time and energy because...they havea million hospitals across the
addressing confidentiality issues raised by administrators, country. But...forthemostpart patient privacy issues for
legal affairs departments, and other entities within insti- us aren’t as muchofan issuebecause wearealready
tutions sharing electronic information. Some spoke of dealingwith that everyday. So...we wereableto
the challenge of having to overcome the resistance, determine...that thedata wasencrypted,that [the
underlying mistrust and ‘aversion’ to sharing data among system] has been vetted and approved byallofthe
players distal to the development of the systems, even differentboards ofeverybody...’—ITstaff
when that sharing was legitimate, lawful, and secure. The
process seemed to be the most cumbersome and
prolonged when academic and governmental agencies were The importance of trust factors in the acceptability of
involved. HIEs
The process for overcoming institutional concerns Trust in the confidentiality and security of electronic
about confidentiality and liability included drafting, data sharing, as well as existing trust among the
differreviewing, re-drafting, and finally approving data-sharing ent stakeholders, had an overarching influence in the
aclegal contractual agreements between collaborating agen- ceptability of the systems. Stakeholders’ personal
cies. Those agreements had to address myriad issues, in- attitudes, expectations about patient perspectives, and
cluding which institution would house the servers for experiences with institutions mutually reinforced one
the HIE, technical discrepancies between IT groups at another. The importance of establishing and maintaining
participating agencies, and the types of information to be trust between patients and providers and between
stakeexchanged. In some cases, the process involved clarifying holders from different institutions was repeatedly
menthe legal parameters of HIPAA, including what kinds of tioned by our interviewees as an issue influencing the
data sharing were forbidden by HIPAA, but also the acceptability of the HIEs. As already noted, a number of
kinds of data sharing activities that were permissible HIE users stated that patient comfort with the HIEs was
without patient consent. For example, at one site, project forged in part by receiving information from the
provistaff had to work with participating institutions to clarify ders. In other cases, stakeholders reflected on how
pathat, under HIPAA, a healthcare plan payer could receive tient trust of the new HIEs was, in a sense, an extension
patient data because it was part of the operations neces- of the trust and satisfaction they felt toward their
persary to coordinate care: sonal providers and the institutions in which they
received care:
‘The case...needs to be driven home to them that
really you can’t use HIPAA to say ‘We can’t give you ‘I think most of our patients feel like their information
that information because of HIPAA..’ HIPAA also is safe. They expect, because of the nature of coming
provides what’s allowed to be shared and under what to a clinic, that it’s based in an academic research
circumstances it’s to be shared. And there institution that there will be information sharing
seems to be an education issue for a lot of within the scope of treatment and coordination of
organizations that are providing care but try to use care and that we will be measuring quality data on
HIPAA... as a weapon to not share data when in them...our patient population at least - is a very
actuality they can and they’re allowed to. But they trusting population. They trust that we’ll keep their
always fall back on,‘We can’t tell you that because of information safe.’—Clinic Coordinator
HIPAA.”—Project Director
Similarly, an administrator at a case management
Stakeholders reported that appeasing concerns of agency mainly serving an ethnic minority referred to the
administrators and legal affairs officers often involved existing trust that clients have in the agency as
overcomdemonstrating that the operations of the HIE would not ing privacy concerns they may otherwise have had:
ultimately reflect poorly on the larger institution.
Although those formal reviews were arduous, they were ‘I think in the communities that we serve there are
successful at assuaging potential concerns and allowing tremendous concerns about privacy, tremendous...
projects to move forward. Similarly, the use of ‘off the bias towards secrecy. Even just as a precaution.... But
shelf’ HIE systems that had been fully vetted by other because most of our staff are themselves [of the same
organizations helped facilitate reviews and approvals at heritage as the patients], because we have a lot of
some of the sites: history with the community, there’s a lot of trust.’Maiorana et al. Implementation Science 2012, 7:34 Page 10 of 14
http://www.implementationscience.com/content/7/1/34
Somewhat different factors influenced how trust be- encryptions, virtual private network (VPN) tunnels, and
tween stakeholders and institutions affected the accept- firewalls. Our interviewees reported that those
explanaability of the HIEs. Within an information exchange tions helped build trust by assuring stakeholders and
system, there are multiple potential dimensions related others involved in decision-making that adequate
safeto trust. Some operate at a within-institution level (e.g., guards would be in place to secure patient data. On the
between an individual provider and the institution where other hand, an IT staff member noted that many
wellhe/she work), and others operate at a between-institu- founded concerns raised by project staff and users not
tions level (e.g., between stakeholders from different ‘technically aware’ could be relatively easily addressed
institutions). In our interviews, the more salient dimen- and explained to them, while IT staff members were
sion was the latter (between institutions), a finding that more concerned about the more complex ‘under the
is likely reflective of the HIE’s fundamental purpose, cover’ system security issues, such as cross-site scripting
which is to facilitate interactions across institutional holes, a vulnerability which allows attackers to bypass
seboundaries. curity mechanisms on web content by injecting
maliIn general, stakeholders from participating agencies cious scripts or codes in order to gain access-privileges
indicated that they trusted the staff and/or the leading to sensitive information:
organization that had taken primary responsibility for
developing each HIE. In most cases, that entity was ei- ‘And it’s a little bit like magic. And you hear a lot of
ther a university and its affiliated hospital or a public things in the press and they’re [stakeholders]
health institution. This confidence extended to the or- concerned with things that have already been taken
ganizing entity’s staff (e.g., technical advisors, IT specia- care of. And the really big technical threats are things
lists, and lawyers) and to the systems infrastructure and that ITstaff are concerned with, we talk about those
security protocols that had been established. As illu- things and follow those things too. There’s always the
strated by the quote below, trust of the developing part- latest exploit or something going on. And we’re trying
ner was particularly salient in interviews conducted at to stay ahead of the typical things that can happen on
collaborating agencies that were small in size or did not the Internet like ....CNS poisoning and all these other
have their own ITsystems in place: things that are going on, to make sure the system can
acquire whatever remedies it has to those things....for
example high level people are worried about‘This chief IT person has been involved in this [name
passwords....Passwords are relatively easy to solve andof the HIE] for a long time. He said ‘we’ve done this
that has to do with access to the system.... Thewith big organizations; this is what we do. ‘.... And the
difference is knowing what can technically happencommunity organizations feel very trusting. I mean
under the covers and seeing what is happening abovethey have this big lawyer IT person coming in saying
the covers...you can have the right password and thethis is what you have to do. They’ve been very
right user ID and have several passwords and all thosetrusting, much more trusting that I expected.... these
sorts of things and still fall a victim to cross sitesmall community based organizations can’t go out and
scripting, and you won’t even know it’s happening.’hire a health information lawyer to go through and
give them an audit and tell them what they need to do
to protect themselves or when they don’t. So when Discussion
they can leverage somebody from [name of the HIE] Because of the stigma still associated with HIV disease,
to come in and say ‘this is the legal way to do it’ they we initially thought that confidentiality might be a key
get that benefit without actually having to expend on barrier to HIE in HIV care settings. Instead, we found
it.’—Project staff widespread acceptability for HIEs across patients,
providers, and other stakeholders. Considering both our large
Across all HIEs, stakeholders indicated that their own quantitative patient dataset and qualitative interviews
trust—and that of the more distal players in positions of with stakeholders at the six sites part of HRSA’s
demonleadership—were driven heavily by confidence in the se- stration project, our findings have broad implications for
curity of the systems. This made IT staff members espe- HIV treatment settings. They suggest that those who
decially key to success not only because these individuals liver and receive HIVcare have grown more comfortable
made the systems operable, but also because their effort’s with integrating HIV-related patient data into modern
quality and thoroughness influenced other stakeholders’ technologies that may improve information flow and
opinions about a HIE’s usability and security. IT staff HIVcare. These attitudes appear to be driven by growing
members’ role was further enhanced by their ability to comfort with information technologies, confidence in the
offer technical support, a task that necessitated clear security protocols utilized to protect data, trust in the
communication to explain in simple language data providers and institutions who use the technologies,