|Home | About | Journals | Submit | Contact Us | Français|
We assessed emergency department (ED) patient acceptance of opt-in, rapid human immunodeficiency virus (HIV) screening and identified demographic characteristics and HIV testing-history factors associated with acceptance of screening.
A random sample of 18- to 55-year-old ED patients was offered rapid HIV screening. Patient acceptance or decline of screening and the reasons for acceptance or decline were analyzed with multivariable regression models. Odds ratios (ORs) with 95% confidence intervals (CIs) were estimated for the logistic regression models.
Of the 2,099 participants, 39.3% accepted HIV screening. In a multinomial regression model, participants who were never married/not partnered, did not have private health insurance, and had 12 or fewer years of education were more likely to be screened due to concern about a possible HIV exposure. In a multivariable logistic regression model, the odds of accepting screening were greater among those who were younger than 40 years old (OR=1.61, 95% CI 1.32, 2.00), nonwhite (OR=1.28, 95% CI 1.04, 1.58), not married (OR=1.82, 95% CI 1.44, 2.28), lacking private health insurance (OR=1.40, 95% CI 1.13, 1.74), and who had 12 or fewer years of education (OR=1.43, 95% CI 1.16, 1.75). Despite use of a standardized protocol, patient acceptance of screening varied by which research assistant asked them to be screened. Patients not previously tested for HIV who were white, married, and 45 years or older and who had private health insurance were more likely to decline HIV screening.
In an opt-in, universal, ED HIV screening program, patient acceptance of screening varied by demography, which indicates that the impact of such screening programs will not be universal. Future research will need to determine methods of increasing uptake of ED HIV screening that transcend patient demographic characteristics, HIV testing history, and motivation for testing.
In 2006, the Centers for Disease Control and Prevention (CDC) advocated for expanded human immunodeficiency virus (HIV) screening in U.S. emergency departments (EDs).1 EDs are desirable locations for HIV screening, as shown by research demonstrating that EDs in some communities provide medical care for large numbers of undiagnosed HIV-infected people2–8 and those at risk of HIV infection.2–4,6,7,9,10 Additionally, many EDs in the U.S. provide care for people without regular sources of medical care who might not get tested for HIV elsewhere.11,12
Although HIV screening in EDs might be a good idea, patient acceptance of ED HIV screening has not been fully elucidated. U.S. ED-based voluntary HIV testing studies have reported a wide range of estimates of the proportion of patients (29.0% to 98.3%) accepting HIV testing in EDs.2,3,5–7,9,13–17 (A summary of these published reports of U.S. ED HIV testing programs from 1993 through 2007 can be found at http://www.brown.edu/Departments/BRUNAP/US_ED_HIV_screening_programs.pdf.) All of these testing programs employed some variation of an opt-in approach in which patients were asked if they would like to be tested, as opposed to an opt-out approach in which patients were informed they would be tested unless they declined.
Variation in acceptance of testing estimates might be explained by differences in protocols across studies, such as whether screening was universal (all patients screened) or targeted (a subset screened based upon risk or demographic characteristics); whether patients who requested testing, providers who referred patients for testing, or patients who were undergoing HIV diagnostic testing were included as those screened; whether incentives to accept screening were offered; when the HIV screening was conducted (24 hours a day vs. selected times and dates); who conducted the screening (clinicians, nurses, medical students, or HIV test counselors/educators who were not a part of the ED staff); what type of HIV test was performed (rapid vs. standard/conventional or phlebotomy vs. finger-stick or oral fluid swab); and differences across studies in the demographic characteristics of the patients screened, the HIV prevalence of the surrounding community, and the exclusion criteria for eligibility for the screening program (e.g., age, recency of testing, language spoken by patient, nature of patient's medical illness, or patient's ability to consent).
Lack of random selection of patients for inclusion in these programs and limited details on how the screening program was conducted make it difficult to estimate the proportion of ED patients in an HIV screening program who would accept testing. An estimate based upon a random sample of patients who are more likely to be screened in a universal, opt-in ED HIV screening program would assist in evaluating the value of ED-based HIV screening and permit future planning for these or other approaches to screening.
Understanding the reasons why ED patients accept or decline screening and the factors associated with accepting or declining screening could also assist in identifying potential barriers to or facilitators of ED HIV screening. Knowing which patients, based upon their demographic characteristics and HIV testing history, are more likely to accept or decline HIV screening would be helpful in developing either tailored or broadly appealing screening programs that might encourage a larger percentage of ED patients to be screened. Furthermore, this information would aid in comparing the value of current opt-in HIV screening programs with future opt-out programs.
We conducted an opt-in, universal, rapid HIV screening program among a random sample of 18- to 55-year-old ED patients who (1) had a subcritical illness or injury, (2) were capable of providing consent for testing, (3) were not referred, and (4) were not provided incentives for testing. For this program, the approach was opt-in (patients were asked if they wanted to be tested), universal (patients were not targeted based upon demographics, risk, or other attributes), and involved screening (instead of diagnostic testing). Acceptance of HIV screening was measured and patients were queried on their reasons for accepting or declining rapid HIV screening in the ED. We also investigated the association of demographic characteristics and HIV testing factors with acceptance of screening and reasons for acceptance or decline of screening.
From July 2005 to July 2006, 18- to 55-year-old ED patients with a subcritical illness or injury were randomly selected for inclusion in an opt-in, universal, rapid HIV screening program study. Participants were surveyed about their acceptance or decline of HIV screening. The hospital Institutional Review Board approved the study.
The study occurred at an academic (provides undergraduate and graduate medical education programs), adult ED in Rhode Island that has more than 95,000 patient visits annually. Approximately 2% of the 18- to 55-year-old ED patients are HIV-infected (Unpublished data, Merchant, November 2007). To be eligible for the study, patients had to be English-speaking, 18 to 55 years old, not presenting for evaluation for a psychiatric illness, not prison inmates, not pregnant, not critically ill or injured, not known to be HIV-infected, not in an HIV vaccine study, and not impaired by a physical or mental disability that would prevent informed consent or participation in the study. Only 18- to 55-year-old patients were included because the state's department of health data indicated that 95% of new HIV infections were occurring in patients of this age group.18 The study was completed prior to CDC's 2006 recommendation for HIV screening until age 64.1
We developed the “HIV testing history and acceptance of rapid HIV testing” questionnaire used for this study by using questions modified from prior studies on HIV and HIV testing, and creating new questions specific to this study; by conducting cognitive-based assessments of the questionnaire through intensive patient interviews of 10 demographically diverse ED patients; and by performing pilot studies of the questionnaire with more than 175 patients. The questionnaire had six demographic questions (age, gender, race/ethnicity, marital status, health insurance status, and education level) and six HIV testing-history questions (history of ever being tested for HIV not as part of a blood donation, history of ever testing positive for HIV, time elapsed since last HIV test, type of last HIV test, main reason for last HIV test, or main reason for never being tested).
The HIV screening acceptance question was stated as follows: “As part of a study, we would like to offer you a free rapid HIV test in the emergency department. This rapid HIV test, which takes about 20 to 30 minutes, involves a finger stick for blood. Would you like to get a rapid HIV test today in the emergency department?” A follow-up question asked participants for the main reason they accepted or declined testing. For the questions asking the main reasons for having been previously tested for HIV, for never having been previously tested for HIV, and for accepting or declining HIV screening, participants were shown a list of potential response options.
Although this was a universal rather than a targeted screening program, a random sample of ED patients was selected for possible inclusion in the study. We conducted a three-stage random selection scheme to obtain a random sample of ED patients that best represented the larger ED population and to avoid any factors involved in patient selection that might produce a biased sample. This scheme entailed randomly selecting (1) the dates on which the study would be conducted each month, (2) the shifts during which the study would be conducted on those dates, and (3) the patients who would be approached for possible inclusion in the study. The dates, shifts, and patients for the study were randomly selected with the assistance of a computerized, random-number-generator program.
The study was eligible to be conducted 24 hours a day, every day (except for eight hospital-mandated holidays). The dates for the study were selected with equal weighting. Shift selection was weighted according to the proportion of patients who presented for medical care to the ED throughout the day; accordingly, 40% of the shifts were from 9 a.m. to 5 p.m., 50% from 5 p.m. to 1 a.m., and 10% from 1 a.m. to 9 a.m. Approximately 30 shifts/month were randomly selected. With the assistance of a computerized, random-number-generator program, we randomly selected a seven-digit sequence of numbers in advance of each shift. These numbers were used to randomly select patients to screen for the study according to the terminal digit of the patients' medical record number. This number is assigned sequentially and permanently to patients when they first enter the hospital system in the inpatient, outpatient, or ED setting. By this method, a random sample of approximately 70% of the patients in the ED during each shift was selected for possible inclusion in the study.
During each shift, research assistants (RAs) conducted a primary eligibility assessment of the ED medical records of patients in the ambulatory care and urgent care areas of the ED who were randomly selected for possible inclusion in the study. The RAs did not conduct these assessments of patients in the critical care and psychiatric care areas of the ED. Patients who appeared to meet study eligibility criteria and who agreed to speak with the RAs underwent in-person, secondary eligibility assessment. During the secondary eligibility assessment, the RAs verified each patient's eligibility for the study. All patients eligible for the study who agreed to be interviewed were asked to participate and were administered the “HIV testing history and acceptance of rapid HIV testing” questionnaire.
No incentives were provided to get patients to participate in the study. The RAs and ED staff were not permitted to encourage or refer patients to be in the study or to be tested for HIV. The RAs followed a strictly defined standardized protocol in the conduct of the study, underwent approximately 40 hours of mock interviews to learn the protocol, and participated in pilot studies to prepare for this study. They were directly observed performing the study twice monthly and were critiqued. Deviations from the protocol were corrected as necessary. The RAs also completed a state-sponsored training program on HIV counseling and testing as well as a training program on rapid HIV testing.
The RAs recorded the study data onto tablet personal computers using a QDS™ CAPI (computer-assisted personal interview) study database.19 The data were entered in duplicate by the RAs with immediate data verification. The data were analyzed using Stata 9.2.20 The results of the medical record screening, eligibility determinations, demographic profiles, and responses to the questionnaire were summarized.
The percentage of participants accepting rapid HIV screening in the ED was calculated for all participants and for each demographic characteristic and HIV testing-history factor. The association of each of these characteristics and factors with acceptance of screening was assessed through bivariate analyses with Pearson's Chi-square testing. The impacts of time waiting in the ED and time of day upon acceptance of HIV screening were examined using Wilcoxon rank-sum and Pearson's Chi-square testing, and logistic regression modeling. A significance level of α=0.05 was used for these and all other analyses.
Multivariable logistic regression analyses were performed to examine the relative relationship of demographic characteristics, HIV testing history, and the influence of the RAs on acceptance of screening. Demographic characteristics and HIV testing-history factors significant in the bivariate analyses were included in the multivariable logistic regression models. Even though the bivariate relationship showed no relationship for gender, gender was included as a covariate in the regression models because prior research showed that a history of HIV testing was related to gender (usually greater for females) and therefore might be related to acceptance of screening.21–23
Three sets of multivariable logistic regression models were formed. The first set of models included all participants in the study, the second set included only those participants previously tested for HIV, and the third set included only those participants who had not previously been tested for HIV. Two separate multivariable models were created for each of these three groups of participants. For each set of models, one model used the demographic characteristics in their categorical form while the second model used them in a dichotomous form. The demographic characteristics and HIV testing-history factors that had the lowest proportion of patients willing to be screened were employed as the reference groups in the models. For the models involving those previously or not previously tested for HIV, covariates were included to examine the relative importance of reasons for previously being tested or not tested for HIV, respectively. Odds ratios (ORs) with 95% confidence intervals (CIs) were estimated.
Multivariable, multinomial logistic regression models were created to determine the relative relationship of the dichotomous form of the demographic characteristics with participant reported reasons for either accepting or declining rapid HIV screening in the ED. The base outcome/reference groups selected for the regression models were the most frequently reported reasons participants gave for either accepting or declining screening. The models compared the odds for which members of a given demographic group chose a reason other than the base outcome for either accepting or declining screening. ORs with 95% CIs were estimated.
Of the 95,000 patients who presented to the ED during the one-year study period for any medical complaint, 14.3% were randomly selected using the three-stage sampling scheme. The Figure depicts the results of the three-stage sampling scheme and the primary and secondary eligibility assessments. Most ED patients not included in the study were ineligible due to age- and language-related exclusions. Approximately 15.0% were not included in the study despite multiple attempts by the RA to approach them for the secondary eligibility assessment. These patients were not available for secondary eligibility assessments because they were undergoing diagnostic testing, examinations, or other procedures, or they were asleep. Another 14.5% would not permit the RA to further assess their eligibility for the study. Table 1 displays the demographic characteristics of the 2,099 study participants. The median age of participants was 32 years (interquartile range: 24 to 42 years); the majority were female and white; and most were single/never married/no partner, had private health insurance, and had completed 12 or more years of formal education.
Of the 2,099 participants, 39.3% accepted rapid HIV screening in the ED. All participants had a negative rapid HIV test. As shown in Table 1, the lowest percentages of participants accepting screening were ages 45 to 49 years, white, and married, with private health insurance and four or more years of college education. In the bivariate analyses, acceptance of screening was associated with each of the demographic characteristics except gender.
Of the 2,099 participants, 54.5% indicated that they had previously been tested for HIV not as part of a blood donation. An additional 22 people stated that they had been tested for HIV as part of a blood donation. Of those previously tested not as part of a blood donation, 42.5% had been tested within the previous year and 97.0% had undergone conventional testing for HIV. As shown in Table 2, acceptance of rapid HIV screening in the ED was lowest among those who previously had not been tested for HIV. In the bivariate analyses, acceptance of screening was related to having ever been tested for HIV, but was not associated with the amount of time elapsed since last being tested for HIV or the type of the participant's last HIV test.
The most common reported reason for a participant's last HIV test was a medical evaluation, and the most common reason for no prior HIV test was a belief of not being at risk for HIV. Among those previously tested, acceptance of ED HIV screening was lowest among those previously required to be tested. Among those not previously tested, acceptance of screening was lowest among those who did not perceive themselves to be at risk for HIV.
Table 3 shows the results of the multivariable logistic regression analyses showing the relative importance of demographic characteristics, HIV testing history, reasons for declining or accepting HIV screening, and the influence of RAs on patient acceptance of rapid HIV screening in the ED. Among all participants, acceptance of ED HIV screening was lower among participants who were older than 40 years, white, and married, and who had private health insurance and more than 12 years of education.
Among those previously tested for HIV, acceptance of ED HIV screening was lower among those who were older than 40 years and married, and those who had more than 12 years of education and who had been required to have an HIV test previously as compared with those who had been tested previously because they wanted to know their HIV status or had a possible HIV exposure. Among those not previously tested for HIV, acceptance was lower among those who were white, married, and had private health insurance, and among those who did not believe themselves to be at risk, as compared with those who had not considered being tested before and those who had not been tested because of fears or concerns about testing. All models also show that despite a standardized protocol and safeguards to ensure that all patients received the same opportunity for screening, acceptance of screening varied by the RA who approached the patient.
The median amount of elapsed time from when participants entered the ED to secondary eligibility determination was 100 minutes. Elapsed time was not related to acceptance of HIV screening by Wilcoxon rank-sum testing (median time: 100 minutes [declined testing] vs. 103 minutes [accepted testing]; p≤0.38) or logistic regression modeling when elapsed time was a linear (p≤0.75) or a quadratic/curvilinear (U-shaped) term (p≤0.37).
The time of day when the patient was approached for screening was also not related to acceptance of screening. By Pearson's Chi-square test, the proportion of participants willing to undergo screening for each of the three eight-hour shifts was the same: 1 a.m. to 9 a.m.—37.8%, 9 a.m. to 5 p.m.—38.7%, and 5 p.m. to 1 a.m.—40.1%; p≤0.75. The time of day when the RA approached the patient was not related to acceptance of screening in a logistic regression model when modeled as a linear (p≤0.33) or a quadratic term (p≤0.37).
As shown in Table 4, the most common reasons patients cited for accepting rapid HIV screening were because they were asked to be tested and because they found it convenient to be tested in the ED. The most common reasons for declining screening were a belief of not being at risk for HIV infection and being distracted by the illness or injury that brought them to the ED. Table 4 also shows the results of the multinomial multivariable logistic regression analyses that explored the relative relationship of demographic characteristics and reasons cited for accepting or declining screening. In regard to accepting screening, participants who were not married, did not have private health insurance, had 12 or fewer years of education, and who had been tested previously for HIV were more likely to state that they wanted to be screened because of their concern about a possible HIV exposure. Participants without private health insurance were more apt to cite convenience as their main reason for agreeing to be screened.
In regard to declining screening, participants who were not married or partnered were more likely to choose another reason for declining screening than a belief that they were not at risk for an infection. Participants who were nonwhite and not married/partnered, and who had previously been tested were more likely to decline HIV screening because they had been tested recently. Males, unmarried participants, and those with 12 or fewer years of formal education more frequently cited not wanting the test or test results. Males and participants who were married were more likely to decline screening because of a time limitation. Nonwhite, unmarried, and previously tested participants cited not wanting a needlestick as the main reason for declining screening.
By the results of this study, we estimate that approximately 39% of English-speaking 18- to 55-year-old ED patients with a subcritical illness or injury would agree to be tested in an opt-in, universal, rapid HIV screening program (subject to the exclusion criteria employed in this study) conducted by nonclinical ED staff. It is difficult to directly compare the proportion accepting testing in this study with previous HIV testing studies, as none of these studies randomly selected patients for inclusion in the study, and details of the testing program methods are incomplete.
Nevertheless, in comparing this study with those shown in our summary of published reports of U.S. ED HIV screening programs, 1993–2007 (available from: URL: http://www.brown.edu/Departments/BRUNAP/US_ED_HIV_screening_programs.pdf), higher acceptance of HIV testing occurred in studies that employed targeted screening, permitted provider referrals for testing, included patients who requested testing, and offered some type of incentive. The estimate from this study can serve as the basis of comparison for future programs that attempt to improve uptake of HIV screening by offering incentives, conducting risk-based targeted screening, using an opt-out approach, or employing other methods. The estimate can also assist ED-based screening programs that use an opt-in approach to anticipate demand for testing and accompanying resources needed. The focus of future research can be to develop methods of increasing uptake of HIV screening in the ED.
The study findings indicate that even when an opt-in screening program is conducted in a universal instead of a targeted fashion, ED patients will not be universally screened. This is an unfortunate observation because one of the hopes for ED-based HIV screening is that it would serve as a means of screening large populations of previously untested people, regardless of their HIV risk. As demonstrated in this study, certain demographic groups will remain less likely to be tested for HIV when an opt-in universal HIV screening program is employed. Even more concerning, some patients who have not previously been tested for HIV will remain untested in an opt-in universal screening program—namely patients who are 45 years and older, white, married, and have private health insurance. These patients form a group for whom other approaches to facilitate screening appear to be necessary.
It is noteworthy that patients who are less likely to ever have been tested for HIV and less likely to accept HIV screening in the ED are part of demographic groups that traditionally have better access to medical care. This finding suggests that access to care does not necessarily confer better access to HIV screening, as it typically does for other types of screening. Universal screening in all health-care settings may help improve HIV screening uptake among these demographic groups. It is encouraging to observe that males who have not been tested elsewhere are at least willing to be screened in the ED. Because males are, in general, less frequent users of routine medical care than females, ED-based HIV screening might be a good testing venue for them. It also is encouraging to note that a third of patients not previously tested for HIV did agree to undergo screening in the ED. Developing ways to increase the percentage of first-time HIV test recipients is an important future goal if the ED is meant to serve as a place to capture those who have not been tested previously.
Because opt-in HIV screening involves an active choice for screening, patient-reported reasons for accepting or declining screening can be helpful in designing future methods to increase the percentage of patients who will be screened in an ED-based program. Wanting to know one's HIV status and having a possible exposure to HIV were associated with a greater acceptance of screening, while declining screening was related to having previously been required to be tested and perceiving oneself to not be at risk for HIV infection. Reasons for acceptance or decline of screening varied by participant demography. In particular, those who were married or not partnered declined HIV screening out of a belief of not being at risk for HIV infection. These findings again suggest that even when screening is universally conducted, uptake will not be universal.
Patients self-select or target themselves for screening based upon their perception of risk and the need for testing. It is likely that years of targeted screening for HIV, based upon risk and demographic characteristics, have helped build a perception that HIV screening is only necessary for those in certain demographic groups and among those at traditional risk for an infection. These findings suggest that a major public health challenge to increasing uptake of HIV screening is to reverse a public perception that HIV testing should be based upon perceived risk. It is not yet known if universal, routine HIV screening, particularly opt-out screening, can help eliminate this perception.
Acceptance of HIV screening also varied according to which RA performed the study. This variation occurred despite patients being randomly selected for the study; the use of a standardized script for how patients were to be asked about HIV screening; extensive training of RAs on the study protocol; and regular, periodic, direct observation and critiques of the RAs. It could be that nonverbal or verbal cues, in the form of subtle encouragement, or unobserved deviations from the protocol could account for this variation. This finding also suggests that interpersonal factors influence testing decisions. Identifying these factors or skills and then teaching or reinforcing them to RAs or staff may facilitate acceptance of screening across populations that remain resistant to screening.
This study had a number of limitations that affected the interpretation of the study results. First, the study was from a single, urban, academic hospital ED in the U.S. HIV testing histories and demographic characteristics are likely to vary widely across the nation's EDs. As such, the patient population included in this study might not be comparable to those in other EDs. However, given the random sampling, ED patient volume-weighted sampling, year-long sampling, and the diverse groups of patients included in the sample, the participants represented a reasonable cross-section of patients at this ED. Given the methods employed in the three-stage sampling scheme, we believe that the 14.3% of ED patients randomly selected for possible inclusion in the study well represent the greater population of English-speaking patients with a subcritical illness or injury who presented for medical care during the study period.
Second, the exclusion criteria and nature of the study limit how the study findings can be generalized. It is not clear if acceptance of ED-based HIV screening using an opt-in approach would be the same for patients who were excluded in this study. However, an opt-in approach to screening might not be appropriate for patients who are not able to provide consent to testing because of their medical illness or injury.
Third, this study did not include an HIV risk-assessment profile, which limited, for example, our ability to determine whether a patient's decision to decline screening was reasonable, given that patient's self-reported risk for HIV. We are currently conducting a study to assess ED patient risk factors for an HIV infection and to determine if patient awareness of their risk for HIV motivates them to be tested in the ED.
Fourth, because this study rested on self-reported data, there might be inaccuracies in objective data points, particularly in the items from the HIV testing history. Social desirability factors might have influenced some patients in their responses to reasons for accepting or declining screening or for having been previously tested or not tested for HIV. Although we cannot objectively assess the extent of the impact of self-reported data upon our analyses, we believe this influence would not have had a significant impact on our findings, given the large sample involved in this study.
Using the random selection process employed in this study, we estimate that EDs should expect approximately 40% of ED patients with a subcritical illness or injury to accept rapid HIV screening in an opt-in, universal, ED screening program. Unfortunately, uptake of screening in this type of program is not universal, which detracts from the intent of ED-based HIV screening as a way to reach large populations of previously untested and at-risk individuals. In this study, patients appeared to self-select for uptake of screening, based upon their demographic characteristics, perception of need for HIV screening, and reasons for previously being tested or not tested for HIV. Motivation to be screened appeared to be greatest among those who previously had a possible exposure to HIV or who wanted to know their HIV status. Declining screening was most often related to patient perception of not being at risk for HIV or previously being required to have an HIV test. Future research appears to be needed to determine methods of improving acceptance of ED HIV screening that transcend variations in uptake based upon demography, HIV testing history, and motivation for screening.
The authors acknowledge the contributions of the following research assistants who made this study possible: Erin Gee, Kelly Cockcroft, Bethany Catanzaro, Erin Argueta, Shereen Mohiuddin, and Jeremy Katzen. The authors also thank Dr. Sheryl Lyss of CDC for her dedication to this project.
The results of this study were presented at the Centers for Disease Control and Prevention (CDC) National HIV Prevention Conference in Atlanta on December 3, 2007.
Dr. Merchant and this study were supported by a career development grant from the National Institute for Allergy and Infectious Diseases (K23 A1060363). The study was also supported by a cooperative agreement grant from CDC (U65/CCU124504). Dr. Mayer was supported by the Center for AIDS Research at Lifespan/Tufts/Brown (P30 AI42853).