HIV SCREENING IN EMERGENCY DEPARTMENTS/CONTEXT
2009 US Emergency Department HIV Testing Practices
Richard E. Rothman, MD, PhD, Yu-Hsiang Hsieh, PhD, Leah Harvey, BA, Samantha Connell, BS, Christopher J. Lindsell,
PhD, Jason Haukoos, MD, MSc, Douglas A. E. White, MD, Aleksandar Kecojevic, MPH, Michael S. Lyons, MD
From the Department of Emergency Medicine, Johns Hopkins University, Baltimore, MD (Rothman, Hsieh, Harvey, Connell, Kecojevic); the Department of
Emergency Medicine, University of Cincinnati, Cincinnati, OH (Lindsell, Lyons); the Department of Emergency Medicine, Denver Health Medical Center,
Denver, CO (Haukoos); and the Department of Emergency Medicine, Alameda County Medical Center, Highland Hospital, Oakland, CA (White).
Objectives: We characterize HIV testing practices and programs in US emergency departments (EDs) in 2009.
Methods: A national Web-based survey of members of the National ED HIV Testing Consortium, participants in
the 2007 Centers for Disease Control and Prevention (CDC)–sponsored ED HIV Testing Workshops, all US
academic EDs, and a weighted random sample of US community EDs with snowball sampling to recruit
additional testing sites was conducted. Data collected included geographic location, estimated seroprevalence,
indications for testing, method of consent, weekly number of tests, funding, and costs.
Results: Of 619 sites surveyed, 338 (54.6%) responded. A total of 277 (82.0%) reported conducting any HIV
testing, and 75 (22.2%) reported systematic HIV testing programs, operationally defined as having testing or
screening organized at the departmental or institutional level. systematic HIV testing programs were
concentrated in the Northeast, at high-volume urban EDs, and in regions with higher HIV/AIDS prevalence. Most
systematic HIV testing programs had existed for less than or equal to 3 years, and nearly one third reported using
an opt-out approach for consent. Among systematic HIV testing programs, the number of patients tested ranged from
less than 1 to 2,100 tests per week. Overall, universal screening was the most commonly reported screening
method reported overall, and rates of HIV positivity were consistently above the CDC threshold of 0.1%.
Conclusion: The number of EDs conducting HIV testing has grown substantially since release of the 2006 CDC
HIV testing recommendations. Although many EDs have systematic HIV testing programs, the majority do not.
Ongoing surveillance will be required to quantify the evolution of ED-based HIV testing and the factors that
facilitate or impede expanded translation. [Ann Emerg Med. 2011;58:S3-S9.]
0196-0644/$-see front matter
Copyright © 2011 by the American College of Emergency Physicians.
The 2006 Centers for Disease Control and Prevention (CDC)
recommendations for HIV testing advanced a strategy whereby
screening for HIV would become part of routine medical practice.1
To facilitate translation of this strategy into action, multiple
operational barriers in previous guidelines, including several
particularly problematic for emergency settings, were either
removed or streamlined.1A survey of academic emergency
departments (EDs) conducted shortly after the release of the 2006
guidelines found that 57% of academic EDs offered rapid testing
under certain circumstances, only 13% recommended routine HIV
screening, and only 4% had adopted routine screening practices.2A
more comprehensive survey that included both academic and a
representative sample of community EDs, conducted in 2007 and
published in this issue,3found that 65% of academic EDs and 50%
of a sample of community EDs performed some type of HIV
testing. Neither of these previous surveys captured information
about the presence of systematic HIV testing programs,
operationally defined here as EDs with testing or screening
organized at the departmental or institutional level.
Since 2006, significant resources have supported ED-based
HIV testing, including a large proportion of the $35 million of
federal funds allocated to expand HIV testing in minority
populations.4The CDC also organized regional strategic
planning workshops to facilitate implementation of HIV
screening in acute care settings, and new legislation has been
passed in many states removing requirements for separate signed
consent for HIV testing.5Further, a National Consortium for
ED HIV Testing has been created, which has informed
implementation strategies,6and the Centers for Medicare &
Medicaid Services has released advisories indicating that HIV
screening, including that in EDs, is reimbursable.7Because of
these initiatives and others, we expected that, as of 2009, the
number of EDs offering HIV testing would have increased.
Previous national surveys about ED HIV testing practice
were limited to mostly academic settings, conducted near or just
after the release of the 2006 recommendations, and did not
define presence or scope of systematic HIV testing programs.
Volume , . : July
Annals of Emergency Medicine S3
Our results provide a crude estimate of the expansion of HIV
testing nationally, describing the extent of systematic HIV
testing programs in both academic and community settings.
Goals of This Investigation
This study was designed to characterize contemporary HIV
testing practices in US EDs in 2009, 3 years after the release of
the 2006 CDC guidelines. Specifically, we sought to describe
the types and locations of EDs performing any testing,
including systematic HIV testing programs, and characterize
these within the broader context of the HIV epidemic.
A cross-sectional Web-based survey was conducted from July 1,
2009, through September 5, 2009. The study was approved by the
Johns Hopkins University Institutional Review Board.
Study Design (Survey Instrument and Outcomes Measures
The survey was designed to elicit information about each
ED’s geographic location, patient demographics, whether HIV
testing was conducted and under what circumstances, where
HIV testing was conducted, the consent process, sources of
funding, and the presence and duration of a systematic HIV
testing program. For EDs with systematic HIV testing programs
we sought additional information on methods of patient
selection for testing and approximate numbers of patients tested
and estimated annual programmatic costs. The survey
instrument was pilot tested to assess feasibility,
comprehensibility, and ease of use. The final survey consisted of
15 questions (Appendix E1, available online at http://www.
annemergmed.com). Nomenclature related to HIV testing and
screening was in accordance with published recommendations.6
Setting and Selection of Participants
We used multiple sources to identify the EDs most likely to
conduct HIV testing and to achieve broad representation of all
geographic regions across the United States. The 4 predetermined
sources were (1) EDs represented at the National ED HIV Testing
Consortium (N?26)6; (2) EDs that participated in the CDC
Strategic Planning Workshops, a series of 2-day workshops held
from 2007 to 2008, intended to engage priority EDs around the
United States to adopt HIV screening (N?195)8; (3) all US
academic EDs, as reported in July 2009 by the Society for
Academic Emergency Medicine (N?139); and (4) a weighted,
randomly generated geographic sample of approximately 20% of
community EDs associated with a hospital with at least 100
inpatient beds (N?418). We also used snowball sampling, a
process whereby all EDs that received a survey were asked to
forward it to, and provide contacts for, other potential respondents
(N?25).9The final sample consisted of 619 sites.
Hospital sites were categorized as an academic site if they
were listed on the Society for Academic Emergency Medicine
Web site as having active emergency medicine residency
training programs, nonacademic teaching sites if they did not
have an active emergency medicine residency training program
but were listed as teaching institutions by the American
Hospital Association, or community hospitals. HIV and AIDS
prevalence data (made available with permission from the
National Minority Quality Forum) were abstracted from the
HIV/AIDS Atlas by the county in which the responding EDs
were located; data were made available with permission from the
National Minority Quality Forum.10The National HIV/AIDS
map10was used to assess geographic distribution of testing sites
in relation to HIV/AIDS prevalence. The number of ED visits
for nonresponding sites was provided by the American Hospital
Association’s Health Research Education Trust (personal
communication, Juliet Yonek, 2010). ED sites were also
categorized according to whether they reported having a
systematic HIV testing program.
Methods of Measurement
The survey was uploaded on Survey Monkey
(SurveyMonkey.com, Portland, OR) and invitations to
participate were initially disseminated by electronic mail on July
2, 2009. Written instructions provided the rationale for the
survey, the link to the instrument, and instructions about
preferred survey respondent, ranked most to least preferred, as
follows: (1) director or program coordinator, if the site had a
systematic HIV testing program; (2) ED director; (3) charge nurse
or nursing director; (4) other ED personnel (eg, social worker).
Each prospective respondent received 2 e-mails. For sites that did
not complete the survey, at least 2 telephone contacts were
attempted and, if contact was established, the survey was conducted
by telephone. Interviews were conducted by trained research
assistants. After initial data analysis, calls were made to verify
apparent outliers and corrections were made as appropriate.
For sites performing no HIV testing, we requested only basic
information about the ED and contact information for other ED
sites at which HIV testing might be performed. For EDs with HIV
testing but no systematic HIV testing program, we requested
responses to an additional set of questions about testing practices
and funding sources. For those with systematic HIV testing
programs, we requested responses for all survey questions.
Primary Data Analysis
We used frequencies and percentages or median and
interquartile range, as appropriate. Associations between
categorical variables were assessed with ?2tests. Data were
managed with Microsoft Excel (Microsoft, Redmond, WA), and
statistical analyses were performed with SAS (version 9.1; SAS
Institute, Inc., Cary, NC).
Of the 619 sites surveyed, 338 (54.6%) responded. Among
them, 75 sites (22.2%) reported systematic HIV testing
programs, 202 (59.7%) reported testing without a systematic
2009 US Emergency Department HIV Testing PracticesRothman et al
S4 Annals of Emergency MedicineVolume , . : July
HIV testing programs, and 61 (18.0%) reported no HIV
testing. Sites that responded did not differ from nonrespondents
by geographic region, but responding sites were more frequently
academic institutions and those with ED volumes greater than
50,000 visits per year. Of respondents, 33 (9.8%) were from
nonacademic teaching hospitals, 96 (28.4%) were from
academic institutions, and 209 (61.8%) were from community
hospitals. Overall, 146 (43.2%) of respondents were recruited
by their participation in the National ED HIV Testing
Consortium or CDC Strategic Workshops, which comprised 40
of the 75 (53.3%) systematic HIV testing programs identified.
Persons completing the survey included ED charge nurses or
nurse managers (64.5%; N?218), HIV testing program
coordinators or program directors (13.3%; N?45), ED
directors or department chairs (12.4 %; N?42), and ED social
workers (0.6%; N?2). The survey was completed anonymously
by 31 respondents (9.2%). An HIV program coordinator or
director completed the survey for 32 of the 75 (42.6%) sites
with systematic HIV testing programs.
Characteristics of EDs that completed the survey are described
in Table 1, stratified by the presence or absence of testing and
whether sites reported having a systematic HIV testing program.
Of the 75 sites reporting systematic HIV testing programs,
85.3% were located in urban settings, 65.3% were academic
institutions, and 79.4% were publicly owned.
Of the 202 sites offering HIV testing without systematic
HIV testing programs, nearly 40% were located in suburban
areas and 74.8% were in community settings. The 61 sites that
reported having no HIV testing tended to be in community
settings with annual volumes of less than 100,000. Of note,
90.5% of hospitals with an ED volume of at least 100,000 visits
per year had HIV testing available (N?38), and more than half
of these reported having a systematic HIV testing program
Characteristics of the HIV testing process for sites that
conduct testing are shown in Table 2, grouped by the presence
or absence of a systematic HIV testing program. For EDs
without systematic HIV testing programs, nearly all tested for
occupational exposures and 74.3% reported physician-initiated
testing, versus EDs with systematic HIV testing programs,
which reported 73.3% occupational exposure testing and 57.3%
physician-initiated testing outside of the systematic HIV testing
program. Opt-in, separate, and written consent methods were
reported most frequently, regardless of whether a systematic
HIV testing program was in place. However, nearly one third of
sites with a systematic HIV testing program reported using an
opt-out approach. Overall, about half of the systematic HIV
testing programs required separate informed consent, but
40.0% indicated that a patient’s signature was not required for
HIV testing. Fewer than 10% of systematic HIV testing
programs had been in operation for more than 5 years; more
than one third had been in operation for less than a year.
Overall, 73 of 338 (21.6%) sites performed HIV screening of
any sort. Testing approaches reported by 75 sites with
systematic HIV testing programs (including 27 sites that
reported using multiple approaches) included universal
Table 1. Descriptive characteristics of the 338 ED sites that completed the survey.*
Presence of HIV Testing, No. (%), N?338
Sites Without STPs,
Sites With STPs,
N?75 (22.2 %)
STP, Systematic HIV testing program.
*Two EDs without systematic HIV testing programs did not reply to the location question and 2 EDs with systematic HIV testing programs did not reply to the owner-
†P values refer to presence of HIV testing for each characteristic.
Rothman et al2009 US Emergency Department HIV Testing Practices
Volume , . : July
Annals of Emergency Medicine S5
screening (42.7%), nontargeted screening (40.0%), targeted
screening (30.7%), and diagnostic testing (34.7%) (Table 3).
Of these 75, 48 (64.0%) indicated that they used only 1
approach for patient selection: 2 (4.2%) diagnostic testing, 19
(39.6%) universal screening, 18 (37.5%) non-targeted
screening, and 9 (18.8%) targeted testing.
Extramural funding for HIV testing was reported by 97 of
the EDs that conducted any HIV testing (35.0%). External
funding was reported by 68 (90.7%) EDs with systematic HIV
testing programs compared with 29 (14.4%) EDs without
systematic HIV testing programs. Sites with systematic HIV
testing programs were most commonly funded by government
grants. Only 25.3% of programs reported explicit research grant
funding. Nearly 20% of sites with systematic HIV testing
programs indicated that some form of HIV testing was billed to
the patient, and 2 systematic HIV testing programs reported
patient billing exclusively. Of sites with systematic HIV testing
programs, 21 (28.0%) estimated their operating costs for
systematic HIV testing, ranging from $10,000 to $325,000 per
year, with a median of $60,000 (interquartile range $40,000 to
For sites with systematic HIV testing programs, the median
estimated number of tests per week varied according to the
testing approach; diagnostic testing was associated with the
lowest number (median 10 tests per week) compared with a
median of 150 per week for sites that reported universal
Table 2. Characteristics of HIV testing at sites that reported conducting any testing.
No. (%), N?202
No. (%), N?75
Situations testing performed*Occupational exposure
Patient’s signature is required
Separate informed consent
State or city health department
Local AIDS administration
Method of consent*
Type of testing program
NA, Not applicable.
†These questions applied only to sites reporting programmatic testing.
‡Met operational definition of “nonprogram” sites; reported recently receiving a program grant.
§External funding from local health departments for sexual assault exposure testing.
Table 3. Characteristics of EDs that reported having systematic testing programs (N?75).
Median Number (Interquartile Range; Range)
Estimated numbers of tests (weekly)
Estimated positivity rate, %*
10 (5-20; 1-100)
2.2 (1-10; 0.08-50)
20 (5-50; 0-100)
1.0 (0.5-3; 0-25)
50 (21-100; 12-300)
1.4 (0.5-2.5; 0.2-25)
150 (78-300; 20-2,100)
0.9 (0.55 -1.55; 0.02-3)
*The numbers of respondents of EDs with eligible estimated positivity rate for diagnostic testing, targeted screening, nontargeted screening, and universal screening pro-
grams were 19, 17, 22, and 20, respectively. Data about tests performed and estimated positivity rates were entirely self-reported and were not independently verified.
2009 US Emergency Department HIV Testing PracticesRothman et al
S6 Annals of Emergency Medicine Volume , . : July
screening (Table 2). The largest number of tests per week was
2,100, about 80% of the ED census, at a site conducting
universal screening. The median number of tests per week at
133 of 202 (65.8%) EDs without systematic HIV testing
programs was 1 (interquartile range 1; range 1 to 150).
Estimated HIV positivity rates were unknown for the majority
(196/202 [97.0%]) of sites without systematic HIV testing
programs. The estimated HIV positivity rate at sites with a
systematic HIV testing program was highest for diagnostic
testing models (median 2.2%) and lowest for those reporting
universal screening (median 0.9%).
EDs located in regions with HIV prevalence above the
CDC-recommended threshold for screening of 0.1% (which
represented approximately 50% of our sample population)
accounted for 74% of those with HIV testing programs (Table
4A). However, only 28.4% of EDs in regions in which the HIV
prevalence was greater than 0.1% reported having a systematic
HIV testing program. Similar associations were observed
between AIDS prevalence and ED HIV testing; nearly 80% of
systematic HIV testing programs were at sites with AIDS
prevalence rates of greater than 0.1% (Table 4B).
The results of this survey are subject to several limitations.
First, although our sample attempted to include all academic
sites and geographically representative nonacademic sites, our
list was not comprehensive. It is also possible that our findings
may be subject to response biases. Specifically, we may have
oversampled hospitals likely predisposed to offer HIV testing
through their participation in either the National ED HIV
Testing Consortium or representation at the CDC Strategic
Planning Workshops. Further, the numbers of sites recruited by
snowball sampling was also much lower than anticipated (25 of
the 619 sites). Accordingly, our results serve principally to
provide a sense of the rapid expansion of ED-based HIV testing
nationally, rather than an accurate representation of the
complete scope of testing currently performed.
The survey was pilot tested by experts in emergency
medicine, infectious disease, and public health, but it is possible
that not all respondents interpreted the survey questions in the
same manner. It is possible that certain elements may have been
confusing. For example, some sites with systematic HIV testing
programs may have failed to report physician-initiated
diagnostic testing, even though the survey instrument explained
that response categories were independent and not mutually
exclusive. In addition, some of the quantitative responses,
especially estimates of prevalence, test volumes, and costs, may
not be entirely accurate. Finally, although we attempted to
reach the most knowledgeable individual at each site about HIV
testing practices, the majority of respondents were nursing
professionals who may have had less knowledge of the details of
their departments’ HIV testing practices than HIV program
coordinators. At sites with systematic HIV testing programs, the
majority of respondents were either physician program directors
or HIV program coordinators.
In this study, we found that 82.0% of participating EDs
offer HIV testing, including 88.5% of academic EDs and
79.4% of nonacademic EDs. Further, we found that 21.6%
(73/338) of participating sites reported that they offer HIV
screening, using a universal, nontargeted, or targeted approach.
These results represent a recent increase in the availability of
HIV testing in US EDs and a steady increase in HIV screening.
In the first national survey of HIV testing in US EDs conducted
in 1996, Wilson et al11reported that slightly more than 50% of
academic EDs tested for HIV under “special” circumstances,
principally occupational and nonoccupational exposure. However,
only 3% were “screened” for HIV, even in situations that clearly
warrant screening, such as with a diagnosis of other sexually
transmitted infections. Since that survey, the CDC has released 2
sets of recommendations. The most recent recommendations
emphasize the pivotal role of US EDs in HIV screening.1,12
Although results of surveys of different target populations are not
Table 4A. Association of ED-based HIV testing program and ED county’s HIV prevalence.*
Proportion of EDs with HIV Testing Program by HIV Prevalence
A, HIV prevalence.
62 (19.3%) 3 (11.5%) 3 (4.6%)10 (14.5%) 10 (15.4%) 22 (34.4%)7 (41.2%)7 (43.8%)
Table 4B. Association of ED-based HIV testing program and ED county’s AIDS prevalence.*
Proportion of EDs with HIV Testing Program by AIDS Prevalence
65 (19.5%)2 (12.5%) 1 (5.6%) 1 (3.4%)6 (9.0%) 25 (19.7%) 16 (31.4%)14 (56.0%)
*Grouped according to the National HIV/AIDS prevalence map.10
†CDC threshold for recommending routine HIV screening for outpatients is 0.1%.
‡HIV and AIDS prevalence data by county were available for 322 and 333 sites, respectively.
Rothman et al 2009 US Emergency Department HIV Testing Practices
Volume , . : July
Annals of Emergency Medicine S7