Personal health records in a public hospital:
experience at the HIV/AIDS clinic at San Francisco
James S Kahn,1Joan F Hilton,2T Van Nunnery,1Skip Leasure,1Kelly M Bryant,1
C Bradley Hare,1David H Thom3
Personal health records (PHRs) are information
repositories; however, PHRs may be less available to
persons in the safety net setting. We deployed a free,
secure, internet-based PHR for persons receiving care at
the AIDS/HIV clinic at San Francisco General Hospital. In
our initial rollout, 221 persons registered for the PHR.
Compared to the entire clinic, these initial users were
more likely to be Caucasian, male, non-Hispanic, on
antiretroviral medications, and have better control of
their HIV infection. The median number of online sessions
was 7 and the median session length was 4 min.
Laboratory results were the most commonly accessed
feature. Patients were satisfied with the PHR and more
than 80% of users agreed that the PHR helped them
manage their medical problems; however, some users
were concerned that their health information was not
accurate or secure. Patients in a safety net setting will
access and use an online PHR.
Personal health records (PHRs) are used to share
health information, increase health understanding,
enhance engagement between patients and their
providers and health organizations, and ultimately
may contribute to improved personal health.1 2The
high degree of interest in PHRs is in part due to the
importance of sharing health information among
organizations and providers.3
eventually may impact health outcomes.5Recently,
consumer-oriented software companies have begun
developing PHR applications, generating significant
interest among health organizations and their
PHRs are a component of health information
technology and are a tool for consumers seeking to
organize their health information.8e11An idealized
PHR should retrieve comprehensive information
from multiple sources and should enable patients to
share their health information.12The information
in a PHR must be secure, portable, and under-
standable, and allow patients to obtain the best
possible advice and services. One concern is that
patients in the safety net setting with poor internet
access and limited health literacy will not have
access to PHRs and thus may not benefit from their
use.13e16We developed a PHR for HIV/AIDS
patients at San Francisco General Hospital (SFGH),
a public hospital, to understand the extent that this
technology would be used and to investigate the
patterns of use.
EXAMPLE AND OBSERVATIONS
myHERO is a secure online PHR (https://myhero.
sfdph.org) established for persons with HIV/AIDS
at UCSF AIDS Clinic at SFGH (screens available in
supplementary material at http://jamia.bmj.com).
Data in the personal health record are automati-
cally populated from HERO (Healthcare Electronic
Record Organizer), the electronic medical record
(EMR) system used in the HIV/AIDS clinic.
Developed by members
Program, HERO is a secure web-based application
that allows clinicians to document and verify
demographics, diagnoses, medications, and labora-
tory data and uses Health Level-7 to exchange
information with outside information systems. All
information entered into HERO is submitted
through a local intranet for storage in a relational
database. Data in myHERO are populated auto-
matically from HERO and include lists of active
(following a 10-day delay to allow clinician review
prior to patient access), drug allergies, scheduled
clinic visits, and tools to promote healthcare
maintenance. Explanations of laboratory data and
normal ranges are provided alongside the laboratory
values listed by clinic visit date. In an effort to
provide the most valuable laboratory values to
patients and after patient and clinician interviews
regarding the laboratory studies of greatest impor-
tance to patients, we decided to display CD4-T
lymphocytes, HIV plasma viral load, hemoglobin,
total cholesterol, high density lipoprotein, low
density lipoprotein, triglycerides, serum creatinine,
serum alanine aminotransferase, and aspartate
aminotransferase. Problem lists and medications are
linked to the National Library of Medicine’s
MedlinePlus (http://medlineplus.gov) and, when
available, images of the medications (jpeg) are
included. Patients may enter their blood pressure,
serum glucose, and weight or record private notes in
myHERO. The application provides online help,
a weekday help desk accessible by phone, and links
to the clinic and to community support and infor-
mation. myHERO was built using the .NET
Framework (version 2.0, Microsoft) a software
development technology that enables developers to
create dynamic web applications. Data from
myHERO users are retrieved from the websites’ log
records and are merged with HERO SQL server
information and extracted for analysis.
A session tracking tool was launched after
myHERO was initiated. This tool measured online
PHR usage via computer log files. The tracking tool
of theSFGH AIDS
<Supplementary data are
published online only at http://
1San Francisco General AIDS
Program, Department of
Medicine, University of California
San Francisco, San Francisco,
California, USA2Department of
Epidemiology and Biostatistics,
University of California San
Francisco, San Francisco,
California, USA3Department of
Family and Community
Medicine, University of California
San Francisco, San Francisco,
James S. Kahn, Ward 84, San
Francisco General Hospital, 995
Potrero Avenue, San Francisco,
California 94110, USA;
Received 21 April 2009
Accepted 12 December 2009
224J Am Med Inform Assoc 2010;17:224e228. doi:10.1136/jamia.2009.000315
recorded session start and end times and the viewed myHERO
pages for each session. In September 2008, a 25-item evaluation
survey was added to myHERO. The survey appeared when users
viewed their PHR. In addition, we sent users three email
requests asking them to complete the survey. No inducement to
complete the survey was offered. The survey was patterned after
commonly used Likert-response satisfaction questionnaires.17 18
It asked users to rate the helpfulness of 13 site features with
response scale of 1¼did not use, 2¼not helpful at all, 3¼a little
helpful, 4¼somewhat helpful, and 5¼very helpful. Five addi-
tional questions asked patients to agree or disagree about the
completeness and accuracy of myHERO information with
a response scale of 1¼do not know, 2¼strongly disagree,
3¼disagree, 4¼agree, and 5¼strongly agree. Patients were also
asked to agree or disagree with statements that myHERO was
easy to access, easy to navigate, supported participation in their
own care by improving management of medical problems,
prepared them for their clinic visit, helped them take charge of
their healthcare, maintained confidentiality, and allowed for
information sharing. Free text comments were also elicited.
Beginning in March 2007, patients with primary care
appointments at the SFGH HIV/AIDS clinic were invited by
their clinicians (physicians and nurse practitioners) to establish
access to myHERO via an online account. As needed, patients
were helped to acquire an email address, taught how to access
the internet, and provided addresses of locations offering free
web access (eg, public libraries). Many patients were unfamiliar
with email. We offered an email account with Yahoo, Gmail,
MSN, and other service providers. We found that patients
selected Yahoo as the preferred email service due to easier to
understand questions for forgotten identification or password.
Remembering passwords and identification for new email users
became a significant issue for some PHR users.
From March 2007 through December 2008, 3760 patients had
a primary care appointment at San Francisco General Hospital’s
AIDS clinic, of whom 221 registered for online access to their
personal health record and 211 activated their account (figure 1).
There was no inducement, other than the availability of the
online information, to influence a person to register, activate, or
view their online personal health record. We considered that
patients with increasing internet involvement would be more
likely to use the system and so we categorized patients in the
clinic into categories with higher levels of recorded technical
proficiency: (1) patients who did not have or did not provide an
email address; (2) patients who provided an email address but did
not register for myHERO; and (3) patients who registered for
myHERO. We contrast the three groups with respect to demo-
graphic characteristics and HIV-related disease characteristics,
and we test for trends across groups using 1-degree-of-freedomc2
tests (ManteleHaenszel test for discrete characteristics; Wald
test for continuous characteristics). For each tracked PHR user,
we define “session count” as the number of distinct PHR sessions;
“months of PHR use” is defined as the difference between the
user’s first and most recent month of access (plus 1); and “usage
rate” as the user’s session count divided by months of PHR use.
We use a Poisson model and Pearson goodness-of-fit statistic
(reported as c2divided by degrees of freedom) to evaluate vari-
ation in the usage rate by months of PHR use and we summarize
the monthly usage rates via medians and interquartile ranges.
Similarly, we summarize the median session duration per patient
via a Poisson model and via medians and interquartile ranges,
overall and by months of PHR use. In addition, we report the
frequency of individually viewed PHR pages. This project was
granted an exemption certification (#0833272) by the University
of California Committee for Human Research (the Investiga-
tional Review Board at UCSF) and was approved by the Privacy
Board of the San Francisco Department of Public Health.
Patients in the clinic were categorized by increasing level of
online access to health information (a surrogate for technical
proficiency) as follows: 3061 clinic patients without or have not
provided an email address, 478 patients with an email address
but did not register for myHERO, and 221 persons registered
users of myHERO (see supplementary material). As online
access increased, patients were younger (less than 40 years old:
29%, 43%, 37%, respectively), more often male (81%, 95%, 95%,
respectively), more often Caucasian (51%, 66%, 78%, respec-
tively), and less often Hispanic (26%, 26%, 15%, respectively).
Although an AIDS diagnosis was equally common among all
subgroups, myHERO users were most likely to be receiving
antiretroviral agents and had better health status in terms of
personal health record users.
Total number of activated
J Am Med Inform Assoc 2010;17:224e228. doi:10.1136/jamia.2009.000315225
lower HIV plasma RNA values (undetectable: 51%, 53%, 65%,
respectively) and CD4 cell counts (below 250 cells/ml: 27%,
20%, 17%, respectively).
Among the 136 patients who used myHERO during the
tracking period, the median months of myHERO use was 6
((25%, 75%)¼(3, 8)), and the median session count was 7 ((25%,
75%)¼(2, 13); (min, max)¼(1, 109)). myHERO usage rate did not
vary with the number of months of PHR use (Poisson mean, 2.0;
95% CI 1.8 to 2.3) sessions/month). The median session length
was 4 min ((25%, 75%)¼(2, 6.5)), and was longer for inexperi-
enced users than for experienced users (figure 2).
In our experience, laboratory results were the most commonly
viewed PHR feature, with 2044 separate visits to the summary
page by 136 users (table 1). The single most commonly viewed
laboratory value was CD4 cell results (891 visits by 110 persons)
and then HIV RNA plasma value results (542 visits by 104
persons). The laboratory values were a source of frustration
because of our policy to publish laboratory values 10 days after
they appear in the EMR. Initially the time frame was 30 days and
moved to 10 days. In our experience and observation, patients
would like the values to be simultaneously populated in the PHR
and the EMR. Office visit lists were viewed 719 times by 117
persons, medical conditions 441 by 107 persons, and medications
366 times by 107 patients. Features for recording weight, blood
pressure, and blood glucose were also utilized. The feature that
allowed patients to enter comments was accessed the least.
Provider profiles were the most accessed clinic information page.
Among the 136 persons for whom we tracked session usage,
51 (38%) completed the survey (table 2). There were no signifi-
cant demographic or HIV-related differences between persons
completing or not completing the survey (data not shown). The
number of missing individual items ranged from 0 to 4. The most
commonly used page contained laboratory values (50; 98% of
respondents), lists of medical conditions (45; 94% of respon-
dents), medication list (42; 88% of respondents), and links to
information about their health conditions (41; 80% of respon-
dents). Links to health conditions, links to community and other
resources, andthe laboratoryresults were reported to be useful by
80e90% of patients. Over two-thirds of patients agreed that the
information provided was complete and accurate, though
a substantial number reported that they believed there were
inaccuracies in current medications (25%) and in medical
conditions (32%). Over 90% of patients agreed that navigation of
myHERO was easy, though somewhat fewer (84%) agreed that
the same was true with regard to accessing the myHERO
website. More than 80% of users agreed that myHERO helped
them manage their medical problems, prepared them better for
their visits, and helped them take charge of their healthcare.
Thirty-one percent did not agree that the information was
confidential. The reasons for the lack of confidentiality were not
captured. Overall, 71% of patients reported being satisfied or very
satisfied with myHERO. Thirty-one comments were provided by
19 users. The most common critical comment was that the
laboratory results were not up to date and the lists of medical
conditions and medications were incomplete or inaccurate.
A secure, portable, and internet-based personal health record was
initiated and used by HIV/AIDS patients receiving care at San
the safety net setting to detail the extent of patient access to the
information in a PHR, including the number of sessions per user,
the duration of the sessions, and patient satisfaction with their
online PHR. We observed a steady increase in the number of
more likely to be male, Caucasian, and non-Hispanic, and to be
receiving antiretroviral medications, to have higher CD4 cells, and
lower HIV viral load; however, the incidence of AIDS was not
email address. We observed that our patients were capable and
desired a PHR to be a convenient tool of information exchange.
patients and Yahoo! tended to be the preferred service provider for
new email users. Forgotten passwords for access to the PHR and
and the most common reason for patients to access our help desk.
Overall, patients found myHERO useful, particularly for
laboratory results, medication lists, medical conditions lists, and
links to resources of information. Reported usefulness was
user and time for each session.
Total number of sessions per
12345678910 11 12 13 14 15 16 18 19 20 21 24 25 26 31 32 36 38 40 41 43 59 110
Average Session Length (Min)
226J Am Med Inform Assoc 2010;17:224e228. doi:10.1136/jamia.2009.000315
somewhat limited by delays of laboratory results populating
myHERO and a surprising number of inaccuracies in medica-
tions and medical conditions. Providing up to date laboratory
results would seem to be especially important, given the popu-
larity of checking lab results. Improving this feature has the
advantage of providing patients with timely information, and
may trigger clinicians to review patients’ laboratory values and
discuss the implications of the laboratory values with their
patients once the values are available. We are now attempting to
change the policy so that laboratory studies that populate the
EMRwill simultaneously populatethe PHR.Patients complained
that their medications and problem lists in myHERO were
inaccurate. MyHERO information is populated from the EMR,
suggesting that in our case, the data in the EMR were inaccurate.
Although the exact nature of the inaccuracies is unclear, our own
chart review would suggest that clinicians fail to update problem
lists and acute problems remain active despite their clinical
resolution. A similar thread applies to inaccurate medications.
Clinicians often fail to discontinue medications that were
administered and were subsequently changed or when the
medication ended such as short course antibiotics. “Medication
reconciliation” is an increasingly recognized element for high-
quality outpatient care.19e21Allowing the patient to review and
verify their information in a PHR and alert their clinician of
discrepancies in their list of conditions and their medication may
help improve the accuracy of their records. Thus we would
recommend more feedback from patients to clinicians to main-
tain accurate information in all electronic systems of healthcare.
Knowing that patients are reviewing their diagnoses and medi-
cations may act as an important stimulus for the clinicians to
maintain accurate information.
LIMITATIONS OF THIS STUDY
We were not able to determine the reasons for persons who
registered an account but did not subsequently activate their
account, although this number was low. We were also not able
to determine why persons with early access did not continue to
log onto their PHR for more sessions. Possible reasons that these
early users did not continue to use their online PHR include
access to the internet, and problems with how to use
a computer or a PHR. There was considerable variation in the
were accessed (N¼136 patients)*
Number of times specific personal health record (PHR) pages
% (no.) of
Personal medical information
Other personal information
Pages for patient to record information
Blood pressure record
Personal notes record
*April 21, 2008 through December 31, 2008.
Results from personal health record (PHR) user survey (N¼51)
Helpfulness of information
List of laboratory results
List of medical conditions
List of current medications
Links to information about conditions
Information on positive health program
Links to community resources
List of allergies
Self-monitor blood pressure
View appointment schedule
Complete and accurate
List of allergies
List of medications
List of medical conditions
Easy to use
Easy to navigate
Easy to access
Allows sharing information
Helps managing medical problems
Makes me better prepared for visits
Helps me take charge
Information is confidential
*Helpful includes “very helpful” and “somewhat helpful”; it excludes “a little helpful” and “not at all helpful”.
yAgreement includes “strongly agree” and “agree”; it excludes “I don’t know”, “disagree”, and “strongly disagree”.
J Am Med Inform Assoc 2010;17:224e228. doi:10.1136/jamia.2009.000315 227
number of sessions and the number of months of PHR use across
individuals. Some users with few months of use were new users,
while others may have had poorer access to the internet, or did
not derive sufficient value from myHERO. Providing better
access to the internet might improve the continuity of use of
a PHR, and a more comprehensive educational service to the
end-users might improve their computer use or increase the
value of the information in myHERO.
A limitation of the satisfaction data is that it is derived from
25% of the initial PHR users and therefore may not be repre-
sentative of all PHR users nor would it represent the larger clinic
population. The small sample size is a limitation of the
preliminary study but still suggests that our patients would use
a PHR. It is possible that number of sessions, pages accessed, and
satisfaction would be different with a more diverse population
with access to a PHR. It is also possible that internet access may
limit PHR use and further work is needed to determine the
relationship between PHR use and internet access.
The main goal of a PHR is to improve patient self-management
that myHERO helped them better manage their health related
problems. The finding that nearly a third of patients did not agree
that the information is confidential is surprising given the
consistent communications between staff and patients regarding
the website, and it is not clear whether this concern with confi-
dentiality impacted patients’ PHR use. It is important in the
future to clarify the issues of confidentiality and PHR use.
Fundamentally, a PHR must provide value to the patient or it will
not be used.22For many patients, myHERO information was
useful; however more applications, more timely information, or
more accurate information will provide patients with greater
utility and will probably lead to greater use of a PHR.8 23e25
In conclusion, this project demonstrated that HIV/AIDS
patients receiving carein a safety netsetting willchoose to create,
activate, and use a PHR. In the future PHRs will be flexible to
changes and improvements, share data, retrieve comprehensive
information from multiple sources, and allow patients or their
advocates to use data to deliver the best possible advice and
services. Patients who adopt PHRs will require that the informa-
tion is protected and private, however data security was
not as dominant a concern. Access to the internet will likely be
a key factor for patients to access their health information
on an online PHR. Patients receiving care from safety net institu-
tions, and vulnerable patient populations, will utilize these new
information technology tools and should have access to online
Funding Dr Kahn acknowledges support from the Commonwealth Fund 20070269;
AHRQ R18 HS017784-01; and the NIH K24RR024369.
Competing interest None.
Ethics approval This project was granted an exemption certification (#0833272) by
the University of California Committee for Human Research (the Investigational Review
Board at UCSF) and was approved by the Privacy Board of the San Francisco
Department of Public Health.
Provenance and peer review Not commissioned; externally peer reviewed.
Kimmel Z, Greenes RA, Liederman E. Personal health records. J Med Pract Manage
Ball MJ, Costin MY, Lehmann C. The personal health record: consumers banking on
their health. Stud Health Technol Inform 2008;134:35e46.
Haux R. Health information systemsdpast, present, future. Int J Med Inf
Ball MJ, Gold J. Banking on health: personal records and information exchange.
J Healthc Inf Manag 2006;20:71e83.
Wuerdeman L, Volk L, Pizziferri L, et al. How accurate is information that patients
contribute to their electronic health record? AMIA Annu Symp Proc 2005:834e8.
Lohr S. Kaiser backs Microsoft patient-data plan. NY Times June 10, 2008.
Steinbrook R. Personally controlled online health datadthe next big thing in medical
care? N Engl J Med 2008;358:1653e6.
Detmer DE, Bloomrosen M, Raymond B, et al. Integrated personal health records:
transformative tools for consumer-centric care. BMC Med Inform Decis Mak
Kupchunas WR. Personal health record: new opportunity for patient education.
Orthop Nurs 2007;26:185e91; [quiz 92e3].
Tang PC, Ash JS, Bates DW, et al. Personal health records: definitions, benefits, and
strategies for overcoming barriers to adoption. J Am Med Inform Assoc
Tang PC, Lansky D. The missing link: bridging the patient-provider health information
gap. Health Aff 2005;24:1290e5.
Kahn JS, Aulakh V, Bosworth A. What it takes: characteristics of the ideal personal
health record. Health Aff (Millwood) 2009;28:369e76.
West DM, Miller EA. The digital divide in public e-health: barriers to accessibility and
privacy in state health department websites. J Health Care Poor Underserved
Graschew G, Roelofs TA, Rakowsky S, et al. Interactive telemedicine as a tool to
avoid a digital divide in the world. Stud Health Technol Inform 2004;103:150e6.
Chang BL, Bakken S, Brown SS, et al. Bridging the digital divide: reaching vulnerable
populations. J Am Med Inform Assoc 2004;11:448e57.
Lorence D, Park H. Web-based consumer health information: public access, digital
division, and remainders. MedGenMed 2006;8:4.
Safran DG, Kosinski M, Tarlov AR, et al. The primary care assessment survey: tests
of data quality and measurement performance. Med Care 1998;36:728e39.
Rubin HR, Gandek B, Rogers WH, et al. Patients’ ratings of outpatient visits in
different practice settings. Results from the medical outcomes study. JAMA
Bernstein L, Frampton J, Minkoff NB, et al. Medication reconciliation: Harvard Pilgrim
Health Care’s approach to improving outpatient medication safety. J Healthc Qual
Dunham DP, Makoul G. Improving medication reconciliation in the 21st century. Curr
Drug Saf 2008;3:227e9.
Orrico KB. Sources and types of discrepancies between electronic medical records
and actual outpatient medication use. J Manag Care Pharm 2008;14:626e31.
Sprague L. Personal health records: the people’s choice? NHPF Issue Brief
Ball MJ, Smith C, Bakalar RS. Personal health records: empowering consumers.
J Healthc Inf Manag 2007;21:76e86.
Raisinghani MS, Young E. Personal health records: key adoption issues and
implications for management. Int J Electron Healthc 2008;4:67e77.
Fanjiang G, von Glahn T, Chang H, et al. Providing patients web-based data to
inform physician choice: if you build it, will they come? J Gen Intern Med
Hysong SJ, Sawhney, Wilson L. Research paper: Provider management strategies of abnormal
test result alerts: a cognitive task analysis. J Am Med Inform Assoc 2010;17:71e77. The third
author’s first name was misspelled and should read Lindsey Wilson. The third author listed in
reference 4 is incorrect. The name should read Vij MS instead of Vilhjalmsson R. Reference 6 is
no longer in press; it was published in September, 2009 (volume 169, issue 17, pages 1578e86).
Reference 18 is no longer in press; it was published in September, 2009 (volume 4: issue 62).
J Am Med Inform Assoc 2010;17:228. doi:10.1197/jamia.M3200corr1
228J Am Med Inform Assoc March 2010 Vol 17 No 2