Outcomes and Cost Analysis of 3 Operational Models for Rapid HIV Testing Services in an Academic Inner-City Emergency Department

Transcription

1 HIV SCREENING IN EMERGENCY DEPARTMENTS/IMPACT/OUTCOMES Outcomes and Cost Analysis of 3 Operational Models for Rapid HIV Testing Services in an Academic Inner-City Emergency Department Yu-Hsiang Hsieh, PhD, MSc, Julianna J. Jung, MD, Judy B. Shahan, RN, MBA, Harold A. Pollack, PhD, Heather S. Hairston, BS, MPH, Daniel Moring-Parris, BA, G. D. Kelen, MD, Richard E. Rothman, MD, PhD From the Department of Emergency Medicine (Hsieh, Jung, Shahan, Hairston, Moring-Parris, Kelen, Rothman) and the Department of Medicine, Division of Infectious Diseases (Rothman), The Johns Hopkins University School of Medicine, Baltimore, MD; and the School of Social Service Administration, Center for Health Administration Studies, University of Chicago, Chicago, IL (Pollack). Objective: We compare the outcomes and costs of alternative staffing models for an emergency department (ED) rapid HIV testing program. Methods: A rapid oral-fluid HIV testing program was instituted in an inner-city ED in Three staffing models were compared during 24.5 months: indigenous medical staff only, exogenous staff only, or exogenous staff plus medical staff (hybrid). Personnel obtained written consent and provided brief pretest counseling, obtained kits, collected specimens, returned specimens to the ED satellite laboratory, and performed posttest counseling and referral to care. Cost analysis was performed to estimate cost per patient tested and cost per patient linked to care. Results: Overall, 44 of 2,958 (1.5%) patients tested received confirmed positive results and 30 (68%) were linked to care. The exogenous staff only model yielded the highest number tested per month (587), and indigenous medical staff only yielded the lowest (57). Significantly higher positivity rates were found in both indigenous medical staff only (2.2%) and hybrid (2.0%) models versus the exogenous staff only model (0.6%) (prevalence rate ratio: 3.7 [95% confidence interval {CI}1.5 to 9.3] versus 3.4 [95% CI 1.5 to 7.8], respectively). All patients with confirmed positive results were linked to care in the indigenous medical staff only model but only approximately 60% were linked to care in the 2 other models (linked to care rate ratio versus exogenous staff only: 1.8 [95% CI 1.1 to 4.4]; versus hybrid: 1.7 [95% CI 1.2 to 2.5]). The indigenous medical staff only model had the highest cost ($109) per patient tested, followed by the hybrid ($87) and the exogenous staff only ($39). However, the indigenous medical staff only model had the lowest cost ($4,937) per patient linked to care, followed by the hybrid ($7,213) and exogenous staff only ($11,454). Conclusion: The exogenous staff only model tested the most patients at the least cost per patient tested. The indigenous medical staff only model identified the fewest patients with unrecognized HIV infection and had the highest cost per patient tested but the lowest cost per patient linked to care. [Ann Emerg Med. 2011;58: S133-S139.] /$-see front matter Copyright 2011 by the American College of Emergency Physicians. doi: /j.annemergmed INTRODUCTION Background For more than 2 decades, researchers at the Johns Hopkins Hospital emergency department (ED) have investigated HIV epidemiology in acute care settings and identified high rates of unrecognized infection in inner-city EDs. 1-9 Studies examined feasibility of different diagnostic assays and explored various HIV ED testing strategies. 4-6,10-13 The overall prevalence of HIV at this ED increased from 5% in the late 1980s to 11% to 12% in 2001 through 2003, but the percentage that was undiagnosed infection decreased from 77% to 18% during that period. 2,6 Two recent studies found rates of newly confirmed HIV positive of approximately 1.5% to 2.5%. 4,5 Importance Since 2001, dozens of ED rapid HIV testing programs have used different testing models and demonstrated the substantial burden of HIV in ED patients. 14 Reported outcomes from these programs have rarely included costs. Recently, Farnham et al 15 developed a model to estimate costs of HIV screening strategies in different health care settings and estimated that the cost per patient receiving HIV test results was lowest for screening in EDs relative to other clinical venues. Only 3 previous studies Annals of Emergency Medicine S133

2 Outcomes and Cost Analysis of 3 Rapid HIV Testing Models have directly reported cost-related outcomes: total program testing costs, cost per patient tested, per patient with confirmed positive results detected, or per patient with confirmed positive results and linked to care Each of these 3 programs used exogenous staff, leaving unresolved the relative costs of using indigenous versus exogenous staff. Cost figures varied, but all suggest that ED-based rapid HIV testing programs are costeffective. Goals of This Investigation The stated goal for the ED testing program under all 3 models was to test all eligible patients for HIV. Our primary objectives were to compare the number and proportion of patients tested, yield of screening, the number of patients linked to care per month, and costs for 3 different staffing models for an ED HIV testing program. Our secondary objective was to compare cost per patient tested, per patient with confirmed positive results detected, and per patient with confirmed positive results and who was linked to care. METHODS Study Design This was a single site retrospective analysis of a natural experiment during which 3 distinct nonsimultaneous staffing models were used for rapid HIV testing. We compared the number of patients tested, yield of screening, and estimated costs (per patient tested, per patient with confirmed positive results detected and linked to care), for each staffing model. Initially the program used indigenous ED staff only. Over time, the program evolved and used 2 other staffing models during phases of different durations. The study was approved by the Johns Hopkins University School of Medicine institutional review board. Setting Johns Hopkins Hospital ED is an academic adult ED of a large, tertiary teaching hospital that primarily serves a socioeconomically disadvantaged inner-city population in Baltimore, MD. Annual ED volume is approximately 60,000 visits; more than 75% of patients are black, HIV prevalence is 10% to 12%, and approximately 15% of patients have ever used injection drugs. 19 Interventions ED Rapid HIV Testing Program. In November 2005, an HIV testing program using the rapid oral fluid OraQuick Advance (OraSure Technologies, Bethlehem, PA) was instituted. All ED patients aged 18 to 64 years were to be offered free rapid HIV testing if they were not critically ill, had no previous diagnosis of HIV or AIDS, reported no HIV test in the previous 3 months, and were able to provide informed consent. Eligibility criteria were the same for all 3 staffing models. We estimated that 3,200 to 3,800 patients were eligible per month. The principal investigator (R.E.R.) was responsible for the design, plan, and training for all 3 staffing models, in Hsieh et al Figure. Timeline of 3 operational staffing models of a rapid HIV testing program implemented in an inner-city academic ED. addition to supervision of the overall testing program. The testing process involved obtaining separate written consent, providing brief pretest counseling (written brochures with abbreviated verbal information), obtaining specimen collection kits, collecting oral mucosal transudate specimens, delivering specimens to the ED satellite laboratory, disclosing test results, and performing posttest counseling. Rapid tests were performed in an on-site ED laboratory. The confirmatory testing procedure included documenting that Western blot was ordered for any patient with reactive rapid test results, arranging for phlebotomist to draw blood, and encouraging patients to remain in the ED until blood was drawn. The process for referral to care varied for the different staffing models. It was designed to be intrinsically consistent with the staffing model, as detailed below. For all 3 models, an ED-based HIV testing program coordinator was responsible for tracking follow-up and assisting after missed follow-up visits (rescheduling the initial appointment and making up to 2 reminder telephone calls after missed appointments). For patients with reactive rapid test results who did not receive a Western blot test in the ED, the program coordinator made 2 follow-up telephone calls asking the patient to return for confirmatory testing. If unsuccessful, the program coordinator provided contact information to the Baltimore City Health Department for follow-up. The sequence of the 3 operational staffing models between November 7, 2005, and December 31, 2007, is illustrated in the Figure. The HIV testing process was performed by indigenous staff (including emergency medicine residents and nurses), HIV testing facilitators (paid staff or volunteer student staff), or both. The timing and duration of each phase were based on convenience. Indigenous Medical Staff Only Model (8 Months; November 2005 to June 2006). The intended mission was to integrate universal HIV screening 20 into ED services by using existing ED staff. Indigenous staff were to offer testing to every eligible patient. This program followed a pilot feasibility project initiated in June 2004 in the immediately adjacent urgent care center. 4 An intensive didactic and role-playing educational program created by 2 of the authors (J.J.J., R.E.R.) immediately preceded the testing program, designed to describe the rationale for and operational features of the planned HIV screening program. 5 The indigenous medical staff only model relied on existing medical staff, mainly residents, with added support from ED S134 Annals of Emergency Medicine

3 Hsieh et al nurses in the last 4 months. For patients with reactive test results, medical staff also called the HIV specialty clinic to book prearranged appointments at the hospital s HIV specialty clinic 1 to 3 weeks after the ED visit for disclosure of confirmatory results and, if confirmed positive, a complete evaluation. Patients with reactive rapid test results but negative Western blot test results were informed by telephone and counseled by the HIV testing coordinator. Exogenous Staff Only Model (2 Months; July to August 2007). Because of resource limitations, the exogenous staff only model was used for only 2 months, during which 13 HIV testing facilitators offered testing to as many eligible patients as possible, 24 hours a day, 7 days a week. All facilitators received state certification in HIV counseling, testing, and referral from the Maryland Department of Health and Mental Hygiene before they started testing. Facilitators determined eligibility by chart review and then approached eligible patients for testing. Medical staff were instructed not to conduct any HIV testing on their own. The HIV testing process was the same as in the indigenous medical staff only model, except for referral: the facilitators (with or without assistance from the HIV program coordinator) called to schedule appointments at the HIV specialty clinic for patients with reactive test results. The program coordinator was responsible for all other aspects of follow-up. Hybrid Staffing Model (14.5 Months; August 2006 to June 2007 and September 2007 to December 2007). HIV testing facilitators (1 to 3 part-time staff) provided coverage during approximately 24% of operating hours. During the remainder, indigenous staff were responsible for HIV testing. Both facilitators and staff and were instructed to offer testing to as many eligible ED patients as possible. All testing processes were the same as described for the indigenous medical staff only model, except for referrals. During weekday working hours (9 AM to 5 PM), the program coordinator was paged to assist medical staff or HIV testing facilitators. During other times, medical staff or HIV testing facilitators followed a written referral process, which was same process that the ED program coordinator used, to schedule appointments at the HIV specialty clinic for patients with reactive test results. The program coordinator disclosed Western blot test results to all patients with reactive rapid test results (positive and indeterminate Western blot test results in person and negative results by telephone). Data Collection and Processing A dedicated program coordinator established a database and maintained daily entries for all 3 staffing models for the numbers of rapid HIV tests performed (collected from the ED satellite laboratory), rapid test results, and, when indicated, Western blot test results and linked to care information obtained from the hospital electronic medical record system. Weekly, the program coordinator reviewed electronic patient records of each confirmed positive case to verify that HIV had not been diagnosed previously. Time spent on each element of Outcomes and Cost Analysis of 3 Rapid HIV Testing Models the rapid testing process was estimated by the program coordinator. Hourly personnel wages (without fringe) and unit costs of rapid HIV tests were obtained from the departmental research administrator. Outcome Measures The primary outcome measures were total numbers and proportions of patients tested, reactive tests, confirmed positive test results, and patients linked to care for each staffing model. A confirmed positive result was defined as a positive Western blot test result after a reactive rapid HIV test. A false-positive rapid HIV test result was defined as a negative Western blot test result after a reactive rapid HIV test; lack of a Western blot test result after a reactive rapid HIV test was considered an unconfirmed result. (This occurred when a patient left the ED before phlebotomy for Western blot test.) Linked to care was considered unsuccessful if there was no documented evidence of a patient with confirmed positive results entering care within 6 months of the initial reactive ED rapid test, despite 2 attempts at follow-up by the program coordinator. Secondary outcome measures were estimated time spent on each element of the testing process and estimated costs per patient tested, per patient with confirmed positive results detected, and per patient with confirmed positive results and linked to care for each model. The overall estimated cost attributable to the testing program (without charges for facilities or space) was the sum of labor, material, and laboratory costs. Labor costs were calculated by multiplying the estimated time spent on each procedure of the HIV testing processes by hourly wages for responsible personnel. Costs were calculated per patient tested, per confirmed HIV-positive patient detected, and per patient linked to care. The actual program cost, an incremental cost to the ED, was operationally defined as total costs of test kits, laboratory fee, and the salaries for exogenous staff, including program coordinator and facilitators, whether or not they were actively engaged in testing. Data Analysis Descriptive statistics, means, and proportions are reported for each staffing model, and primary outcomes were normalized as outcomes per month to account for the different durations of each model. Rate ratios and confidence intervals (CIs) were calculated to assess the significance of differences between HIV positivity rates and linked to care rates for each model. Exact CIs for the ratio of 2 proportions were estimated if 25% of the cells had expected counts less than 5. All statistical analyses were performed with SAS (version 9.1; SAS Institute, Inc., Cary, NC) or StatXact-8 (Cytel Inc., Cambridge, MA). RESULTS Overall, 2,958 ED patients received rapid HIV tests during the 24.5-month study period. Sixty-six (2.2%) patients tested reactive on the rapid test and 44 (1.5%) were confirmed as being newly diagnosed HIV positive. Of the remaining 22 Annals of Emergency Medicine S135

4 Outcomes and Cost Analysis of 3 Rapid HIV Testing Models Table 1. A, Materials and personnel costs in an ED-based rapid HIV testing service program. Items Units Cost/Hourly Wage, $* Laboratory testing HIV test kit cost Laboratory flat fee 4.04 Personnel Principal investigator compensation ED nurses ED residents Program coordinator Facilitator hourly compensation Facilities ED facilities/space No charge *Hourly wage did not contain fringe. B, Crude estimated time spent in hour in each procedure of an EDbased rapid HIV testing service program. Items Time Spent, h Procedures Pretest counseling 0.08 Consent for HIV testing 0.08 Obtain test kit* 0.17 Collect specimen 0.02 Return specimen to ED satellite laboratory 0.02 Posttest counseling nonreactive 0.03 Posttest counseling reactive 0.50 Arrange confirmative Western blot test 0.33 Arrange referral to HIV care 0.03 Follow-up on linkage to care 0.17 *Includes time spent on traveling back and forth to ED satellite laboratory, documenting, labeling, and waiting to pick up the test kit. Includes time spent on arranging phlebotomy, documenting test order, and persuading patients not to leave the ED before blood was drawn. Includes time spent on making a telephone call to the Johns Hopkins Hospital HIV clinic for appointments. patients results, 12 were false positive, 1 was indeterminate, and 9 were unconfirmed. Thirty (68.2%) patients with confirmed positive results were linked to care. Materials and personnel costs are summarized in Table 1A and the crude estimated time spent on each procedure is presented in Table 1B. The estimated total program costs (based on all staff costs attributable to HIV testing) were $49,377 for the indigenous medical staff only model, $45,747 for the exogenous staff only model, and $115,884 for the hybrid model. The actual program costs for each model (test kits, laboratory fee, and exogenous staff) were $32,903 for the indigenous medical staff only model, $37,775 for the exogenous staff only model, and $81,617 for the hybrid model. The total number of patients tested and cost per patient tested, found to have reactive results, found to have confirmed positive results, and linked to care by each model are summarized in Table 2. Per month, the exogenous staff only model had the highest number of patients tested, reactive and confirmed positive test results, and patients with confirmed positive results and linked to care, the indigenous medical staff Table 2. Outcomes and cost analysis results of 3 operational staffing models in an ED-based rapid HIV testing program. Characteristics Hsieh et al Operational Staffing Models (Months) IMSO ESO Hybrid (8) (2) (14.5) Total patients tested 453 1,173 1,332 Patients tested/mo Rapid test result: reactive, No. (%) 15 (3.3) 13 (1.1) 38 (2.9) Reactive/mo Reference test result: confirmed, 10 (2.2) 7 (0.6) 27 (2.0) No. (%) Confirmed/mo Not confirmed False positive Unknown Indeterminate Linked to care (of confirmed 10 (100) 4 (57) 16 (59) cases), No. (%) Linked/mo Cost per patient tested, $ Cost per reactive patient detected, $ 3,291 3,524 3,037 Cost per patient with confirmed 4,937 6,545 4,292 positive results detected, $ Cost per patient linked to care, $ 4,937 11,454 7,213 only model had the lowest. A significantly higher HIV positivity rate was observed in the indigenous medical staff only (2.2%) and hybrid (2.0%) models versus the exogenous staff only (0.6%) model (prevalence rate ratio: 3.7 [95% CI 1.5 to 9.3] and 3.4 [95% CI 1.5 to 7.8], respectively). All 10 confirmed HIV-positive patients in the indigenous medical staff only model were linked to care compared with 20 (59%) of the 34 patients with confirmed positive results in the 2 other models (linked to care rate ratio: 1.8 [95% CI 1.1 to 4.4] for indigenous medical staff only versus exogenous staff only; 1.7 [95% CI 1.2 to 2.5] for indigenous medical staff only versus hybrid). False positivity was highest in the exogenous staff only model (27%) and unconfirmed reactive rapid test results highest in the indigenous medical staff only model (20%), but differences by staffing model were not statistically significant. The indigenous medical staff only model had the highest estimated cost per patient tested ($109), followed by the hybrid models ($87) and the exogenous staff only model ($39). The exogenous staff only model screened the largest number of patients and identified the largest number of HIV patients per month but had the highest cost per confirmed HIV-positive patient detected ($6,545), followed by the indigenous medical staff only model ($4,937) and the hybrid model ($4,292). The exogenous staff only model also had the highest cost per patient linked to care ($11,454), followed by the hybrid model ($7,213) and indigenous medical staff only model ($4,937). LIMITATIONS There are several limitations in our study. First, the study was not a concurrent, blinded, randomized controlled trial but a S136 Annals of Emergency Medicine

5 Hsieh et al retrospective review of an operational program. Therefore, certain staffing models may have promoted testing more vigorously than others. Second, the staffing models were used during different periods and the exogenous staff only was the only one conducted during the sometimes chaotic period during changeover to new house staff, so we cannot rule out potential observer and temporal biases. Third, the duration of each model was very different: 8 months for the indigenous medical staff only model, 2 months for exogenous staff only model, and 14.5 months for the hybrid model. Because of its much shorter duration, outcomes from the exogenous staff only model may have been more subject to the influence of startup considerations (which might have reduced productivity) or by enthusiastic new staff (which might have increased productivity). Fourth, labor costs were estimated without observational data for actual time spent on each component of the testing process or for cost allocations when more than 1 type of personnel could perform the same function. 17 Thus, the cost comparisons among the staffing models may not be accurate. Fifth, our study likely suffers from lack of external validity. This is a single-institution study. Operational staffing models that we used in our ED may not be suitable for other EDs, and thus results may not be generalized. Personnel costs, laboratory costs, and time spent on each testing procedure are also likely institution specific. Sixth, all rapid tests in our study were performed in the ED satellite laboratory, unlike that in most other published reports, in which either indigenous or exogenous staff performed the rapid tests themselves. The ability of the laboratory to accommodate a limited number of tests per hour restricted the number of patients who could be tested, especially during ED peak hours, independent of the staffing model. Seventh, modification in the referral procedures under different staffing models might have had a major effect on the number of patients linked to care, which could have confounded observed outcomes attributable to the staffing model. Finally, unmeasured factors (eg, stage of disease or need for admission of HIV-infected patients identified) may also have contributed to differential rates of linked to care and biased the calculated cost per patient with confirmed positive results and linked to care. DISCUSSION The outcomes from our HIV testing program are comparable to those from similar HIV testing programs in adult EDs since The numbers of patients tested per month (57 for indigenous staff, 587 for exogenous staff, 92 for the hybrid model) are within the wide range reported by others (23 to 857) ,21-29 Although each of the 3 staffing models was intended to reach all eligible ED patients for HIV testing, we did not achieve universal screening in any of the phases. The exogenous staff only approach appeared to follow a universal testing, compared with others, in which the percentage of positive patients suggests that targeted screening and diagnostic testing predominated. Outcomes and Cost Analysis of 3 Rapid HIV Testing Models We found only 2 other indigenous staff models documented in the literature. Both were Centers for Disease Control and Prevention funded demonstration projects, and each adopted different operational approaches to testing, yielding significant differences in the monthly numbers of patients tested (23 in Denver and 473 in Oakland, CA). 24,28 In Denver, emergency physicians were instructed to identify high-risk patients who were tested by clinical social workers, but Oakland used a mixed model with aggressive promotion of universal screening by triage nurses, supplemented by physician-ordered diagnostic testing. Our indigenous medical staff only model was intended to offer testing to all eligible ED patients but tested significantly lower numbers of patients per month than the program in Oakland. The low numbers tested and high positivity rates suggest that medical staff selected high-risk or symptomatic patients according to clinical gestalt. To date, exogenous staffing models have been used in the majority of ED-based HIV testing programs, with differing intensity of coverage. For example, George Washington University 18 used undergraduate students for 3 months, with 16 hours of coverage per day, and tested 857 patients per month. However, average numbers decreased to approximately 350 per month in a subsequent report of a 15-month operation. 22 The number of patients tested in our 2-month exogenous staff only model was similar, with 587 tested per month. Our hybrid model (with only 24% coverage of ED operating hours) produced results similar to those of other programs. 16,17,21,23,,25,29,30 Differences in yield might also arise from the need for separate written consent for HIV testing, location of testing (bedside versus laboratory), and rates of patients previously identified as being HIV positive in the ED population. Our overall confirmed HIV seropositivity rate, 1.5%, was higher than the 1.0% estimated from a previous report from 2001 to 2003 from our site 6 and represents one of the highest reported rates of newly recognized HIV infections from an ED. The 0.6% positivity rate observed with exogenous staff only is comparable to that reported by other studies that intended universal screening. 18,28 Previous studies have consistently demonstrated that diagnostic or targeted testing yields much higher rates of confirmed HIV positivity but lower numbers of HIV-positive patients. 25,28 Results from our study, ie, increased average number of confirmed positive results per month during periods when more patients were tested, support those observations. The number of unrecognized infected patients identified during a normalized period (per month or per year) as a measure to compare ED-based HIV screening programs is not commonly used in the literature. Using this indicator allowed us to compare the yields of our 3 different staffing models and provides a different, but important, metric for yield compared with positivity rate. The exogenous staff only model had the highest number of patients with confirmed positive results identified per month and the indigenous medical staff only Annals of Emergency Medicine S137

6 Outcomes and Cost Analysis of 3 Rapid HIV Testing Models model had the lowest, the inverse of the positivity rates. From a public health perspective, the number of confirmed positive results identified per unit of time is a better indicator to determine how well the program is able to identify undiagnosed infection in individuals. Caution must be applied when comparing one program data to another, with consideration of the total hours of coverage. An alternative approach to test more eligible patients in the ED would be to perform a conventional HIV test on all patients having blood drawn for other reasons. 1,6,31 Because 50% to 60% of patients have blood drawn during their visit in our ED, the number of patients tested and number of unrecognized HIV infections would likely increase severalfold. This approach might be particularly applicable to EDs such as ours in which testing is performed in the laboratory. In the early 2000s, several ED HIV testing programs implemented experienced poor rates of linked to care, some as low as 30%. 17,30 More recent linked to care rates range from 79% to 89%. 18,21,23,24,28 Our indigenous medical staff only model had 100% linked to care, but the exogenous staff only and hybrid models successfully linked only about 60% of patients. One possible explanation of the difference in linkage rates between the indigenous medical staff only model and the other 2 staffing models (which had lower but similar rates) is the person responsible for the initial referral in each model (medical provider versus testing facilitator). Alternate explanations that were not ascertained might include differences in patient characteristics, such as stage of disease at diagnosis, medical insurance or lack thereof, housing status (eg, homeless), or other psychosocial factors. Regardless of the staffing model, only between 2 and 7 patients per month with reactive rapid test results required follow-up, a workload that should be manageable with any staff model. Only 3 recent ED rapid HIV testing programs report cost per patient tested, per patient with confirmed positive results identified, and per patient with confirmed positive results and linked to care. Our estimated cost per patient tested was slightly higher than that observed by others, but our costs per confirmed positive result and per patient linked to care were similar However, unlike those in other studies, staff in our program did not perform the actual rapid test. Our cost per case detected (and those in the 3 other studies ) are less than those for breast or cervical cancer screening, for which costs per case detected are $10,566 and $13,340, respectively. 32 CONCLUSION Of the 3 staffing models we evaluated, the model that relied on exogenous staff exhibited the lowest cost per patient tested, but the models that used indigenous staff (indigenous medical staff only and hybrid) yielded lower cost per patient with confirmed positive results detected and linked to care. Our cost analysis provides a comparison of relative costs between staffing models and not incremental costeffectiveness indicators that directly compare advantages of one staffing model over another. 33 EDs that intend to initiate an HIV screening program must consider, in addition to costs, site-specific factors such as degree of support from the medical staff, operational flow, funding sources, coverage hours, and laboratory capability when designing a program most appropriate for their ED. Any opinions, findings, conclusions, or recommendations expressed in this publication are those of the authors and do not represent the policy or position of the Maryland Department of Health and Mental Hygiene and the Maryland AIDS Administration. Funding and support: By Annals policy, all authors are required to disclose any and all commercial, financial, and other relationships in any way related to the subject of this article as per ICMJE conflict of interest guidelines (see All authors certify no potential conflicts of interest. Supported in part by the Maryland Department of Health and Mental Hygiene and the Maryland AIDS Administration by a Centers for Disease Control and Prevention grant. Supplementary support for expanded HIV testing was funded by grants from The Gilead Foundation. Presented at the Society for Academic Emergency Medicine annual meeting, May 2009, New Orleans, LA. Publication of this article was supported by Centers for Disease Control and Prevention, Atlanta, GA. Hsieh et al Address for correspondence: Yu-Hsiang Hsieh, PhD, MSc, Johns Hopkins University Department of Emergency Medicine, 5801 Smith Ave, Ste 3220 Davis Bldg, Baltimore, MD 21209; , fax ; yhsieh1@jhmi.edu. REFERENCES 1. Kelen G, DiGiovanna T, Bisson L, et al. Human immunodeficiency virus infection in emergency department patients. Epidemiology, clinical presentations, and risk to health care workers: the Johns Hopkins experience. JAMA. 1989;262: Kelen G, Fritz S, Qaqish B, et al. Unrecognized human immunodeficiency virus infection in emergency department patients. N Engl J Med. 1988;318: Kelen G. Human immunodeficiency virus and the emergency department: risks and risk protection for health care providers. Ann Emerg Med. 1990;19: Hsieh Y-H, Dunn E, Codori N, et al. Evaluation and outcomes of inner city urgent care rapid point-of-care HIV testing. Paper presented at: the 16th Biennial Meeting of the International Society for Sexually Transmitted Disease Research (ISSTDR); July 11-13, 2005; Amsterdam, Netherlands. 5. Hsieh Y-H, Jung J, Shahan J, et al. Emergency medicine resident attitudes and perceptions of HIV testing before and after a focused training program and testing implementation. Acad Emerg Med. 2009;16: Laeyendecker O, Rothman R, Henson C, et al. The effect of viral suppression on cross-sectional incidence testing in the Johns Hopkins Hospital emergency department. J Acquir Immune Defic Syndr. 2008;48: Kelen G, Shahan J, Rothman R, et al. Long-term trends in HIV infection in an academic medical center ED. Acad Emerg Med. 2002;9: Kelen G, Hexter D, Hansen K, et al. Trends in human immunodeficiency virus (HIV) infection among a patient population S138 Annals of Emergency Medicine

7 Hsieh et al Outcomes and Cost Analysis of 3 Rapid HIV Testing Models of an inner-city emergency department: implications for emergency department based screening programs for HIV infection. Clin Infect Dis. 1995;21: Kelen G, Johnson G, DiGiovanna T, et al. Profile of patients with human immunodeficiency virus infection presenting to an innercity emergency department: preliminary report. Ann Emerg Med. 1990;19: Kelen G, Bennecoff T, Kline R, et al. Evaluation of two rapid screening assays for the detection of human immunodeficiency virus-1 infection in emergency department patients. Am J Emerg Med. 1991;9: Kelen G, Chanmugam A, Meyer WR, et al. Detection of HIV-1 by polymerase chain reaction and culture in seronegative intravenous drug users in an inner-city emergency department. Ann Emerg Med. 1993;22: Kelen G, Hexter D, Hansen K, et al. Feasibility of an emergency department based, risk-targeted voluntary HIV screening program. Ann Emerg Med. 1996;27: Kelen G, Shahan J, Quinn T. Emergency department based HIV screening and counseling: experience with rapid and standard serologic testing. Ann Emerg Med. 1999;33: Kelen G, Rothman R. Emergency department based HIV testing: too little, but not too late. Ann Emerg Med. 2009;54: Farnham P, Hutchinson A, Sansom S, et al. Comparing the costs of HIV screening strategies and technologies in health-care settings. Public Health Rep. 2008;123(suppl 3): Mehta S, Hall J, Greenwald J, et al. Patient risks, outcomes, and costs of voluntary HIV testing at five testing sites within a medical center. Public Health Rep. 2008;123: Silva A, Glick N, Lyss S, et al. Implementing an HIV and sexually transmitted disease screening program in an emergency department. Ann Emerg Med. 2007;49: Brown J, Shesser R, Simon G, et al. Routine HIV screening in the emergency department using the new US Centers for Disease Control and Prevention guidelines: results from a high-prevalence area. J Acquir Immune Defic Syndr. 2007;46: Hsieh Y, Rothman R, Bartlett J, et al. HIV seropositivity predicts longer duration of stay and rehospitalization among nonbacteremic febrile injection drug users with skin and soft tissue infections. J Acquir Immune Defic Syndr. 2008;49: Lyons M, Lindsell C, Haukoos J, et al. Nomenclature and definitions for emergency department human immunodeficiency virus (HIV) testing: report from the 2007 conference of the National Emergency Department HIV Testing Consortium. Acad Emerg Med. 2009;16: Calderon Y, Leider J, Hailpern S, et al. High-volume rapid HIV testing in an urban emergency department. AIDS Patient Care STDS. 2009;23: Brown J, Kuo I, Bellows J, et al. Patient perceptions and acceptance of routine emergency department HIV testing. Public Health Rep. 2008;123(suppl 3): Centers for Disease Control and Prevention (CDC). Rapid HIV testing in emergency departments three US sites, January 2005 March MMWR Morb Mortal Wkly Rep. 2007;56: Haukoos J, Hopkins E, Eliopoulos V, et al. Development and implementation of a model to improve identification of patients infected with HIV using diagnostic rapid testing in the emergency department. Acad Emerg Med. 2007;14: Lyss S, Branson B, Kroc K, et al. Detecting unsuspected HIV infection with a rapid whole-blood HIV test in an urban emergency department. J Acquir Immune Defic Syndr. 2007;44: Merchant R, Seage G, Mayer K, et al. Emergency department patient acceptance of opt-in, universal, rapid HIV screening. Public Health Rep. 2008;123(suppl 3): Walensky R, Arbelaez C, Reichmann W, et al. Revising expectations from rapid HIV tests in the emergency department. Ann Intern Med. 2008;149: White D, Scribner A, Schulden J, et al. Results of a rapid HIV screening and diagnostic testing program in an urban emergency department. Ann Emerg Med. 2009;54: Kendrick S, Kroc K, Couture E, et al. Comparison of point-of-care rapid HIV testing in three clinical venues. AIDS. 2004;18: Mehta S, Hall J, Lyss S, et al. Adult and pediatric emergency department sexually transmitted disease and HIV screening: programmatic overview and outcomes. Acad Emerg Med. 2007; 14: McCarthy JJ, Green P, Malone K, et al. A novel approach to emergency department HIV testing conventional testing rapidly: 427 new positives and 55,529 tests in 16 months. Presented at: 2010 Society for Academic Emergency Medicine annual meeting; June 3-6, 2010; Phoenix, AZ. 32. Ekwueme D, Gardner J, Subramanian S, et al. Cost analysis of the National Breast and Cervical Cancer Early Detection Program: selected states, 2003 to Cancer. 2008;112: Ekwueme D, Pinkerton S, Holtgrave D, et al. Cost comparison of three HIV counseling and testing technologies. Am J Prev Med. 2003;25: Annals of Emergency Medicine S139