Acceptance of Rapid HIV Screening in a Southeastern Emergency Department

Authors

  • Arin E. Freeman MPH,

    1. From the Department of Emergency Medicine (AF, RWS, JAW) and the Department of Biostatistics (KMM, JKD), School of Medicine, Medical College of Georgia, Augusta, GA.
    Search for more papers by this author
  • Richard W. Sattin MD,

    1. From the Department of Emergency Medicine (AF, RWS, JAW) and the Department of Biostatistics (KMM, JKD), School of Medicine, Medical College of Georgia, Augusta, GA.
    Search for more papers by this author
  • Kelly M. Miller MS,

    1. From the Department of Emergency Medicine (AF, RWS, JAW) and the Department of Biostatistics (KMM, JKD), School of Medicine, Medical College of Georgia, Augusta, GA.
    Search for more papers by this author
  • James K. Dias PhD,

    1. From the Department of Emergency Medicine (AF, RWS, JAW) and the Department of Biostatistics (KMM, JKD), School of Medicine, Medical College of Georgia, Augusta, GA.
    Search for more papers by this author
  • James A. Wilde MD

    1. From the Department of Emergency Medicine (AF, RWS, JAW) and the Department of Biostatistics (KMM, JKD), School of Medicine, Medical College of Georgia, Augusta, GA.
    Search for more papers by this author

  • This study was funded by the Georgia Department of Human Resources Contract 427-93-009090681-99 through the Centers for Disease Control and Prevention Grant 1U62PS000789-01.

Address for correspondence and reprints: Arin Freeman, MPH; e-mail: arfreeman@mcg.edu.

Abstract

Objectives:  The objective was to assess the acceptance of an emergency department (ED) human immunodeficiency virus (HIV) screening program based on the Centers for Disease Control and Prevention (CDC) recommendations for routine HIV screening in health care settings.

Methods:  Rapid HIV screening was offered on an opt-out basis to patients aged 13 to 64 years presenting to the ED by trained HIV counselors. Patients were excluded if they had a history of HIV, were physically or mentally incapacitated, did not understand their right to opt-out, or did not speak English or Spanish. Statistical analyses, including logistic regression, were performed to assess the associations between the demographics of patients offered testing and their test acceptance or refusal.

Results:  From March 2008 to January 2009, a total of 5,080 (91%) of the 5,585 patients offered the HIV test accepted, and 506 (9%) refused. White and married patients were less likely to accept testing than those who were African American and unmarried (p < 0.001). Adult patients were almost twice as likely to accept testing as pediatric patients (odds ratio [OR] = 1.95; 95% confidence interval [CI] = 1.50 to 2.53). As age increased among pediatric patients, testing refusal decreased (OR = 0.71; 95% CI = 0.59 to 0.85), and as age increased among adult patients, testing refusal increased (OR = 1.17; 95% CI = 1.12 to 1.22). Two percent of persons accepting the test were considered high risk. Males were more likely to report high-risk behavior than females (OR = 1.83; 95% CI = 1.23 to 2.72).

Conclusions:  The opt-out approach results in high acceptance of routine HIV screening. Widespread adoption of the CDC’s recommendations, although feasible, will require significant increases in resources.

More than 1 million Americans are infected with the human immunodeficiency virus (HIV), but approximately 21% of these people are unaware of their infection,1 and those unaware account for 54% to 70% of new infections.2 Persons who are unaware of their HIV infection are less likely to reduce behaviors associated with HIV transmission than those who are aware of their status.3 Emergency departments (EDs) not only provide health services for the acutely injured and ill, but also provide rapid access to health care for persons irregularly entering the health system, including the underserved and uninsured. Persons with undiagnosed HIV infection access the ED prior to their diagnosis, but are not often tested.4–6 The ED can, therefore, play an important public health role by offering HIV testing to its patients.

Several EDs have implemented risk-targeted HIV screening programs in which HIV testing is offered based on behavioral risks or clinical symptoms.7–10 HIV infection is, however, increasing among populations that were not considered high-risk populations, such as young people, women, racial and ethnic minorities, persons residing outside metropolitan areas, and heterosexual men and women who are frequently unaware of their HIV risk.11 Thus, in 2006, the Centers for Disease Control and Prevention (CDC) changed their previous recommendation on risk-based HIV testing in health care settings. The CDC now recommends that “opt-out” HIV screening for all patients aged 13–64 years be a part of routine clinical care in health care settings.12 Opt-out screening occurs when the patient is informed either verbally or in writing that an HIV test is going to be conducted unless the patient declines testing.12

Acceptance of routine HIV screening among patients is important in avoiding “missed opportunities” to diagnose unknown HIV infection. The average acceptance rate in EDs performing routine screening is about 51% using screening models that required separate consent for HIV testing or for study participation, not for opt-out screening.13–19 The opt-out approach has been found to be more effective in both prenatal and sexually transmitted disease services than traditional opt-in screening,20–26 but it is unclear how patients will accept opt-out HIV screening in the ED setting. A cross-sectional survey revealed that 81% of patients surveyed would have agreed to testing using the opt-out method.27 The one published study in which opt-out screening was actually used showed a 53% acceptance rate.28

Evaluation of how patients will accept HIV testing is critical in determining if recommendations by the CDC for routine screening in the ED are practical. In this article we report results from an ongoing rapid HIV screening program in a Southeastern ED that assesses acceptance rates by selected demographic characteristics.

Methods

Study Design

Between March 2008 and January 2009, rapid HIV screening was offered, on an opt-out basis, to patients aged 13 to 64 years presenting to the ED. This project was approved by the institutional review board of the Medical College of Georgia.

Study Setting and Population

This project was performed in the Medical College of Georgia ED, a public, academic ED and Level 1 Trauma Center in Augusta, Georgia, with approximately 75,000 annual visits. Augusta is located in east central Georgia, on the border with South Carolina. About 56% of all ED visits are by African Americans and 56% are by females. Approximately 49,500 annual visits (66%) are among persons ages 13 to 64 years. The seroprevalence rate for the area was approximately 0.4% in 2006. Seroprevalence was computed as the number of cumulative HIV/AIDS cases in Georgia’s East Central Health District and two bordering South Carolina counties (Aiken and Edgefield counties), reported in 2006, divided by the census population estimates for 2006.

All persons aged 13 to 64 years who presented to the ED for an illness or injury during a routine testing period were eligible to be tested by trained HIV counselors. Patients were excluded from testing if they opted out, had a history of HIV, were physically or mentally incapacitated, did not understand their right to opt-out, or did not speak English or Spanish. Other patients were not tested if they were discharged prior to being approached, unable to be located, or in public areas within the ED (i.e., hallway).

Study Protocol

Rapid HIV testing was offered between 8am to 6pm, Monday through Friday. Patients who accepted were tested with an oral swab using the OraQuick rapid HIV-1/2 antibody test (OraSure Technologies, Inc., Bethlehem, PA). The manufacturer of OraQuick rapid HIV-1/2 antibody test indicates a sensitivity of 99.3% (95% confidence interval [CI] = 98.4% to 99.7%) and a specificity of 99.8% (95% CI = 99.6% to 99.9%) with oral fluid.29 Preliminary positive results suggest that HIV antibodies may be present and must be confirmed, while a negative result means no HIV antibodies were detected. According to Georgia’s HIV testing policies, written consent for HIV testing is not required; however, the person to be tested must be counseled before the test and have the opportunity to refuse the test. Brochures approved by the Georgia Department of Human Resources that are provided to the person to be tested are considered to have met the requirement for counseling. Triage nurses provided information related to possible HIV testing and the ability to opt-out during the ED registration process, and specific written consent was not required for the rapid HIV test to be performed.

HIV counseling and screening were performed by persons experienced in conducting these services in community settings, as part of the Medical College of Georgia Ryan White Outreach program. They used patient registration forms, which included basic demographics, to identify patients fitting the eligibility criteria. If patients met eligibility requirements, the HIV counselors used the ED tracking system to locate the patient in the ED. Patients were either approached in the exam rooms or were escorted from the waiting room into a private room. The HIV counselors informed eligible patients that an HIV test would be performed unless they declined. If patients did not opt-out of testing, HIV counselors gathered information on patient risk behaviors and provided literacy and culturally appropriate information brochures that included basic information about HIV/AIDS, how the OraQuick Advance rapid HIV-1/2 antibody test worked, and what the test results mean. After answering questions from patients about the test, the HIV counselors performed the test. Patients were given their test results while they were still in the ED. High-risk negative patients were referred to HIV prevention counseling services. For patients aged 13–17 years accompanied by a parent, the test was offered in the presence of the parent; both the parent and the patient had to not opt-out of testing, and parents were asked to leave the room when risk information was obtained.

If the rapid test was preliminary positive, patients aged 18 years or older were notified of their result and received posttest counseling. A blood sample was obtained via venipuncture during their ED visit to confirm the rapid HIV test results with a Western blot test. If the patient was 13 to 17 years old and was accompanied by parent or legal guardian, blood for Western blot testing was collected during their ED visit, and both the teen and the parent or guardian received posttest counseling. If the patient was 13 to 17 years and was not accompanied by a parent or legal guardian and was not an emancipated minor, the patient was told that the test was uncertain and an appointment was made in the pediatric infectious disease clinic or adolescent clinic for follow-up with the patient in the presence of their parent or legal guardian.

Western blot results were given in person by the HIV counselors within 1 week for all adult patients having a preliminary positive rapid HIV test. The HIV counselor established a time and a location within the hospital, usually the ED, to give the patient his or her Western blot result. If the patient was confirmed positive, the HIV counselor offered follow-up treatment in the Medical College of Georgia adult infectious disease clinic. If the patient accepted treatment, the HIV counselor scheduled the initial clinic appointment and accompanied the patient to the first clinic visit. If the patient was 13 to 17 years of age, all arrangements to give Western blot result and for follow-up treatment, if confirmed, were made by the pediatric infectious disease clinic. The local health department attempted to locate all confirmed positive patients who were lost to follow-up for their Western blot result appointment or their initial clinic appointment.

A standardized data collection form was used to collect patient data that were obtained from the patient’s registration form and from self-report. The registration form was used to collect the following information for all patients who registered in the ED during testing hours: name, medical record number, age, race, sex, date of birth, marital status, county of residence, state of residence, and chief complaint for current visit. The specific criteria that excluded patients who were not tested were recorded. Acceptance or refusal of HIV testing was recorded for all eligible patients who were reached. Patient HIV testing histories, HIV risk behaviors, and test results were recorded for all eligible patients who accepted the test (did not opt-out). Information related to confirmatory testing for all patients with preliminary positive test results was documented. Treatment and referral data were collected on all patients who were confirmed positive.

Data Analysis

The data were input into a secure database and analyzed using SAS Version 9.1.3. (SAS Institute, Inc., Cary, NC). Standard descriptive statistics were calculated. We also calculated the number eligible to be screened, number screened, reasons for nonscreening, number of indeterminate rapid HIV tests, number of positive rapid HIV tests, and number of confirmed HIV tests via Western blot. The chi-square test was used to assess the association between whether an eligible patient was offered HIV testing and various demographics, and the association between patient acceptance of the HIV test and selected demographic variables. The Cochran-Mantel-Haenszel (CMH) test was used to assess the association between age, categorized as adult (age 18–64 years) or pediatric (age 13–17 years) and whether the patient accepted HIV testing while controlling for both race and sex. Logistic regression analysis was used to assess the joint effect of various demographic variables on the odds of accepting the test for both adult and pediatric patients. Patients were considered at high risk for HIV if they reported one or more of the following behaviors: sex with a intravenous drug user, sex with a male who had sex with a male (MSM) (for females), sex with a person known to be HIV positive, use of intravenous drugs, use of intravenous drugs and shared equipment, or any MSM during the past 12 months. Logistic regression analysis was used to determine the effect of various demographic variables on HIV risk status (i.e., high risk or not high risk). Unlike linear regression, it is not typical to provide R2 values for logistic regression models. We computed the Hosmer-Lemeshow goodness-of-fit test to give an indication of the fit for the three models. Large p-values indicate a good fit and small p-values indicate a poor fit.

Results

From March 2008 to January 2009, there were a total of 11,120 age-eligible patients presenting to the ED during routine screening periods (Figure 1). Of these, 3,749 were considered ineligible for the following reasons: tested elsewhere within the past year (35.8%), physically unstable (24.4%), tested in the current program within the past year (22.8%), mentally unstable (9.9%), previously diagnosed with HIV (6.3%), unable to speak English or Spanish (0.6%), and dying in the ED (0.2%). Of the remaining 7,371 eligible for testing, HIV counselors approached 5,585 (75.7%). Reasons for eligible patients not being approached were patients being discharged before tested (15.2%), patients not being in a private area (4.9%), and the inability to locate patients within the ED (4.1%).

Figure 1.

 Results of HIV screening, March 2008 through January 2009. Description of eligible patients approached or not approached, eligible patients accepting the test or refusing the test, and the test results of patients tested and the reasons for ineligibility of patients meeting age requirements presenting to the ED during screening hours.

Of the 5,585 eligible patients approached during routine screening periods, 5,080 (91%) accepted testing and were tested (Figure 1). Three of the 5,080 tests were false positive and 5,051 were negative, resulting in a false-positive rate of 0.06% and a specificity of 99.94%. The positive predictive value of the test was 88.9% (24 true positives divided by [24 true positives and 3 false positives]). Of the 28 patients with preliminary positive results, 26 (93%) received their Western Blot results. Twenty-one (88%) of the confirmed positive patients attended their initial clinic appointment.

Among the patients accepting testing, 3,007 (59.2%) were African American and 1,904 (37.5%) were white. Fifty-nine percent (n = 3,011) of patients tested were female and 2,108 (41%) were male. The pediatric population represented 8.6% (n = 435) of patients tested.

Of the eligible males, 74% were approached, whereas 77% of the females were approached. Thus, males were less likely to be approached for the HIV screening (p < 0.01). No other demographic factors were significantly associated with being approached for testing.

Overall, acceptance was high regardless of demographic factor (Table 1). However, pediatric patients, whites, and married patients were significantly less likely to accept the test than older patients, African American, and unmarried patients (p < 0.001 for each variable). Acceptance rates were similar for sex and state of residence (p = 0.603 and p = 0.760, respectively)

Table 1. 
Association Between HIV Tests Occurring and Various Demographic Variables
 TestedRejected Testp-value
  1. *Significant at the 0.05 level. Values are presented as n (%)

ED
 Adult (18–64 yr)4,645 (92)426 (8)<0.001*
 Pediatric (13–17 yr)435 (85)79 (15)
Age, yr
 13–17435 (85)79 (15)<0.001*
 18–342,258 (94)134 (6)
 35–541,873 (90)215 (10)
 55–64514 (87)77 (13)
Race
 White1,904 (89)239 (11)<0.001*
 African American3,007 (92)247 (8)
 Other164 (89)20 (11)
Sex
 Male2,068 (91)200 (9)0.603
 Female3,011 (91)306 (9)
State of residence
 Georgia4,108 (91)402 (9)0.760
 South Carolina918 (90)98 (10)
 Other48 (91)5 (9)
Marital status
 Married1,128 (88)160 (12)<0.001*
 Single3,123 (92)268 (8)
 Widowed90 (92)8 (8)
 Divorced475 (91)45 (9)
 Unmarried with other14 (100)0 (0)
 Separated244 (91)25 (9)
Total5,080 (91)506 (9) 

There was a significant association between age and acceptance of the test after controlling for race and sex (CMH, p < 0.001; data not shown). Pediatric patients were almost twice as likely to refuse testing than adults (odds ratio [OR] = 1.95, 95% CI = 1.50 to 2.53). The Breslow-Day test for homogeneity of the ORs was nonsignificant (p = 0.108). Thus, the ORs did not show statistically significant effect modification for accepting testing among the six race by sex by age group categories.

Adult patients aged 18–34, 35–54, and 55–64 years had acceptance rates of 94, 89, and 87%, respectively. Logistic regression analyses revealed that only age had a significant effect on accepting or not accepting the test. For every 5-year increase in age, the odds that an adult patient did not accept testing increased 17% (OR = 1.17; 95% CI = 1.12 to 1.22; Table 2). Among pediatric patients, for every 1-year increase in age, the odds that a pediatric patient did not accept the test decreased by 29% (OR = 0.71; 95% CI = 0.59 to 0.85; Table 3). Pediatric patients age 13 and 14 years had acceptance rates of 68% and 81%, respectively, whereas pediatric patients aged 15 to 17 years had acceptance rates ranging from 87% to 90%. The logistic regression models used to assess the effects of certain demographic variables on test acceptance for adult patients and pediatric patients had large p-values that exceeded 0.05 (0.099 and 0.433, respectively), indicating a good fit.

Table 2. 
Logistic Regression Results for Adult Patients for Whether They Refuse HIV Testing
DemographicUnitOR95% CIp-value*
  1. *Hosmer-Lemeshow.

  2. †Significant at the 0.05 level.

Sex (male vs. female)10.84(0.68–1.04)0.099
Race (white vs. other)11.04(0.59–1.85)
Race (African American vs. other)10.67(0.38–1.19)
Age, yr51.17(1.12–1.22)†
Table 3. 
Logistic Regression Results for Pediatric Patients for Whether They Refuse HIV Testing
DemographicUnitOR95% CIp-value*
  1. *Hosmer-Lemeshow.

  2. †Significant at the 0.05 level.

Sex (male vs. female)11.47(0.89–2.41)0.433
Race (white vs. other)10.68(0.23–2.07)
Race (African American vs. other)10.35(0.12–1.04)
Age, yr10.71(0.59–0.85)†

Of persons accepting the test, 112 (2%) were considered high risk, based on the previously described definition, while 98% of persons accepting the test either did not report high-risk behavior or did not complete a risk assessment. When assessing the effects of selected demographic variables on high-risk status, age and race were not significant predictors of high-risk behavior, but males were 83% more likely to report high-risk behavior than females (OR = 1.83; 95% CI = 1.23 to 2.72). Among the males tested, 3% were high risk whereas 2% of the females were high risk. The logistic regression model used to assess the effect of certain demographic variables on high-risk status also had a large p-value of 0.097, indicating a good fit.

Discussion

Our results show that opt-out screening for HIV in an ED with trained counselors can lead to very high patient acceptance rates among persons age 13 to 64 years. Our findings suggest that age may be the most predictive factor of test acceptance. The youngest pediatric patients had the lowest acceptance rates, but as their age increased so did testing acceptance. Our findings also show that the rapid HIV test has extremely high specificity with few false-positive results.

Comparison of acceptance rates of HIV screening among studies can be problematic due to differences in testing models and procedures. Nonetheless, the acceptance rate in our program is much higher than that noted in most testing programs that have offered routine HIV screening.13–19,28,30 Although many of these routine HIV screening programs demonstrated that over half of the patients tested will accept testing, most used screening models that required separate consent for HIV testing or for study participation and did not use the opt-out approach currently recommended by the CDC. Our findings illustrate that streamlining the consent and counseling procedures and the opt-out approach for HIV testing in health care settings may promote high test acceptance in the ED. Many states have policies that require separate informed consent and extensive pretest counseling for HIV testing, making implementation of a routine screening program based on the CDC recommendations unpractical.31 Our program demonstrates that persons will readily accept HIV testing without extensive consent and counseling procedures.

Our finding that pediatric patients were less likely than adults to accept testing contradicts other findings, where acceptance among pediatric patients was higher than adults.16 Other routine testing studies that included both adult and adolescent patients found that younger age groups were more likely to agree to test,30 while others found no difference in acceptance by age.18 Most routine HIV testing programs did not include testing the pediatric population.13–15,17,32 However, of the studies that tested pediatric patients, few have tested individuals as young as age 13 years, who had the lowest acceptance rate in our program. Because our pediatric patients were often accompanied by a parent, it is possible that parental influence affected test acceptance among pediatric patients. Furthermore, low perception of risk has been cited as a reason for HIV testing refusal among adolescents.33 Thus, our finding that testing acceptance among pediatric patients increased with age is logical, because HIV risk behavior and risk perception are also likely to increase with age.

Although adolescents were less likely to accept testing than adults, the 85% acceptance rate observed in our program is much higher than in other recent studies offering routine HIV screening to adolescents.16,34 In addition to low perception of risk, never having been offered a test has been identified as a reason for not having an HIV test among adolescents.33 Moreover, ED-based testing programs with extensive counseling and consent procedures seem to reduce the percentage of adolescents tested.34 In 2006, persons age 13 to 29 years accounted for the largest number of new HIV infections (34%).35 Furthermore, the Youth Risk Behavior Survey (YRBS) indicated that almost 50% of adolescents have had sexual intercourse, and of sexually active adolescents, 40% did not use a condom in their previous sexual encounter.36 Our findings demonstrating the overall high acceptability of HIV screening among adolescents suggest that routine HIV screening in the ED is feasible and may provide the opportunity to identify unknown HIV infection in this patient population. Additional efforts may be necessary to increase acceptance among younger adolescents and their parents.

African Americans had similar acceptance rates to whites, which is an important finding because new cases of HIV are occurring in African Americans at a much higher rate than other racial groups.35 Other studies have found that African Americans are likely to accept routine testing in ED and urgent care settings.14,17,18,28,30,37 However, persons tested late in the course of their infection are more likely to be African American or Hispanic,38 indicating that these populations may have more barriers to accessing or seeking HIV testing. African Americans residing in the south are disproportionately affected by HIV/AIDS.39,40 Our high acceptance rate among African Americans, especially in a southern ED where undiagnosed HIV may be most prevalent, suggests that routine testing in the ED may promote earlier diagnosis among this population.

Early detection of HIV infection is important so that interventions, such as highly active antiretroviral therapy (HAART), may be undertaken to prevent further transmission and to reduce the high costs of caring for a patient with active AIDS. Furthermore, research shows that HIV-infected persons aware of their status reduce behaviors associated with HIV transmission.3 Thus, HIV diagnoses early in disease progression will not only improve quality of life and life expectancy of infected persons, but will also reduce the incidence of HIV. Similar to other ED-based routine screening programs, our program identified persons with unknown HIV infection that were likely to leave the ED undiagnosed if not offered an HIV test.

Brown et al.28 also performed an HIV screening program based on the CDC recommendations for testing in health care settings, with an acceptance rate of HIV screening of 53%, which is not appreciably higher than the average acceptance (51%) of other routine screening programs using the opt-in methodology13–19 and much lower than ours. Thus, in addition to the use of the opt-out approach, there are several other features of our program that seem to contribute to the higher acceptance rates.

Like our program, Brown et al.28 and many other programs have used outside staff to perform HIV screening;13–17,30 however, many of our counselors had at least 5 years of experience providing HIV counseling and testing in outreach programs through the Medical College of Georgia. The inclusion of personnel with previous HIV counseling and testing experience may play an important role in the development and implementation of a program having high acceptance of testing. Our HIV counselors were included in the program planning activities and developed the procedures for approaching patients by adapting the counseling and testing methods used in outreach settings to fit the ED setting. High levels of competency were also achieved by our inexperienced HIV counselors through their participation in an extensive training program, which provided them with skills to offer patients the test in a nonthreatening manner, perform pre- and posttest counseling, link patients to care, and perform the technical and quality assurance activities of the OraQuick rapid test.

Routine testing programs have offered HIV testing at various time points of a patient’s ED visit, some offering the test at triage.13,19 Our HIV counselors offered testing mostly in patient exam rooms, which patients may find to be more private than other areas of the ED. Furthermore, to be integrated as members of the ED staff, HIV counselors wore lab coats. We suspect that their professional appearance helped to improve their credibility and gain patient trust.

HIV counselors used a standard script when offering the test to patients. This script emphasized several factors, including that the test is offered to everyone within the testing periods and is free, and the results will be provided to the patient prior to leaving the ED. The script also briefly described HIV transmission modes, which possibly assisted patients with determining their risk. This may be a critical component, because low-risk perception has commonly been cited as a reason for patients declining routine HIV screening.15,17,28,37 HIV counselors also had the knowledge to elaborate beyond the script to address patients’ questions or concerns. Finally, emphasizing that we were implementing the recommendations of the CDC, the nation’s public health agency, possibly helped establish the legitimacy of our program with patients.

Our HIV counselors were successful at providing brief yet thorough verbal pretest counseling in addition to an informational brochure meeting Georgia’s minimum counseling requirements while not interfering with the patient flow. Acceptance of the program and counselors by physicians, nurses, and other health care providers in the ED may also have increased acceptance by patients, through allowing counselors to provide the additional pretest counseling that was likely to influence patient acceptance.

The HIV counselors were representative of the races of our dominant patient populations, with half of them being African American and half being white. Testing was not offered by matching the race of the HIV counselor and the patient. Given our high overall acceptance, we do not believe racial concordance to be a major factor in patient acceptance, but rather the knowledge and skill level of our HIV counselors.

Recent publications have reported a low specificity of OraQuick with use of oral fluid and as a result have modified testing procedures.41,42 These recent reports have caused widespread concern with the performance of the OraQuick test. Our program had a low false-positive rate (0.06%), which is consistent with the manufacturer’s reported specificity. Our ongoing quality assurance activities with the HIV counselors seem to maintain the test’s performance standards as specified by the manufacturer.

Our program was limited because we only tested patients 5 days a week, 10 hours a day. Of the total 47,000 age-eligible visits to the ED that would have been available if we had 24/7/365 screening, we approached about 11%. This finding is fairly consistent with other programs using supplemental staff, which offered HIV screening to less than 10% of the overall ED census.13,15,18 Many may argue, therefore, that use of existing ED staff is the best strategy for maximizing the proportion of ED patients offered testing. While our existing ED staff was supportive of HIV testing during routine testing hours, repeated efforts to train ED health professionals to test persons presenting to the ED with sexually transmitted diseases or suspected AIDS-defining illnesses were not successful. Only 73 patients were tested by ED staff outside of routine testing hours, which supports other findings illustrating that reliance on existing ED staff may result in low levels of testing.8 Furthermore, studies using existing staff to perform testing have lower percentages of persons accepting testing actually tested due to the time constraints, limited staff, and other priorities.13,19

Of those completing the risk assessment, few (2%–3%) were considered at high risk for infection. However, studies have shown that many people with HIV infection do not perceive themselves as being at risk.43–45 Thus, testing based on risk assessment may be less effective than routine testing in identifying unknown HIV infection.46,47 Moreover, many of our ineligible patients were tested within the past year, with a large percentage tested within our screening program on a previous visit. The CDC recommends testing persons at high risk for HIV at least annually.12 Although we used a 1-year period between ED visit and last HIV test to avoid repeat testing of frequent visitors, there are not any clear recommendations on how often routine testing should be performed on persons not meeting high-risk criteria or not meeting any clinical criteria for testing.

Limitations

Routine HIV screening was conducted for only 8- to 10-hour periods on weekdays randomly between the hours of 8am and 8pm, and thus our results might be subject to sampling bias. We conducted a week pilot period of testing from 8am to 9pm, which revealed that patients arriving at night were more likely to be ineligible due to being either physically or mentally unstable. Nonetheless, there may have been differences in acceptance rates among any persons not meeting eligibility criteria during or outside of our normal screening periods. Self-report was used to assess behavioral risk and is, therefore, subject to measurement bias, since persons often underreport HIV risk behavior.48,49 Also, behavioral risk factor data on persons opting-out or ineligible were not collected, so comparing these factors between persons accepting testing with those not tested was not possible. Persons with high-risk behaviors may have been less likely to participate, although anecdotally there was no less participation among high-risk persons. The high overall acceptance rate suggests that all persons were likely to accept testing, regardless of their risk status. Data on the reasons why patients decided to opt-out were not collected, so understanding how to improve our acceptance rates is limited. Although several approaches were used to improve acceptance by patients (e.g., opt-out testing, use of experienced HIV counselors as staff, and cooperation of existing ED staff), data were not collected to determine which component of our screening strategy had the largest impact on patient acceptance. Finally, external funding was available to carry out this project in our ED and to hire staff. Without similar funding, changes in practice, and changes in insurance practices, it is unlikely that such a screening process could be instituted widely in EDs.

Conclusions

Routine rapid HIV testing in patients presenting to our ED using opt-out testing resulted in a 91% acceptance rate among those offered testing. A number of new cases of HIV were detected that would have been missed if routine testing was not available. Widespread adoption of the CDC’s recommendations on screening all patients age 13 to 64 years in health care facilities, although feasible, will require significant increases in resources.

The authors thank Larry Howell for grant administration and for facilitating the collaboration between our project and the Ryan White Outreach program; Sowjanya Mohan, MD, and Mary Forsyth, MD, for providing assistance with linking patients to care in the infectious disease clinic; Joyce Oliver and Martha Ponder for HIV test quality control supervision and performing Western blot testing; Deborah Ivins for HIV counselor training; Kathleen Childs, David Thompson, Shawn Walker, and Tim Wilson for performing HIV screening; Lee Cromwell, Anh Le, Robby Powell, and Courtney Wimmer for data entry; and Bryan Ihrig for assistance with preparation of the manuscript.

Ancillary