Payer Status, Race/Ethnicity, and Acceptance of Free Routine Opt-Out Rapid HIV Screening Among Emergency Department Patients

Abstract

Objectives. We estimated associations between payer status, race/ethnicity, and acceptance of nontargeted opt-out rapid HIV screening in the emergency department (ED).

Methods. We analyzed data from a prospective clinical trial between 2007 and 2009 at Denver Health. Patients in the ED were offered free HIV testing. Patient demographics and payer status were collected, and we used multivariable logistic regression to estimate associations with HIV testing acceptance.

Results. A total of 31 525 patients made 44 765 unique visits: 40% were White, 37% Hispanic, 14% Black, 1% Asian, and 7% unknown race/ethnicity. Of all visits, 10 237 (23%) agreed to HIV testing; 27% were self-pay, 23% state-sponsored, 18% Medicaid, 13% commercial insurance, 12% Medicare, and 8% another payer source. Compared with commercial insurance patients, self-pay patients (odds ratio [OR] = 1.63; 95% confidence interval [CI] = 1.51, 1.75), state-sponsored patients (OR = 1.64; 95% CI = 1.52, 1.77), and Medicaid patients (OR = 1.24; 95% CI = 1.14, 1.34) had increased odds of accepting testing. Compared with White patients, Black (OR = 1.29; 95% CI = 1.21, 1.38) and Hispanic (OR = 1.17; 95% CI = 1.11, 1.23) patients had increased odds of accepting testing.

Conclusions. Many ED patients are uninsured or subsidized through government programs and are more likely to consent to free rapid HIV testing.

Despite substantial public health efforts, infection with HIV remains an important cause of preventable death in the United States.¹ It is estimated that 230 000 people remain unaware of their infections in the United States and 56 300 new infections occur each year, most of which are attributable to contact with those who remain unaware of their HIV status.^2–5

In an effort to have a further impact on the epidemiology of HIV infection in the United States, the Centers for Disease Control and Prevention (CDC) published revised recommendations for HIV testing in health care settings in 2006.⁶ These recommendations attempted to reduce exceptionalism associated with HIV testing by, in part, advocating the performance of routine (nontargeted) screening with an opt-out consent approach. Unfortunately, at that time, little was known about the effectiveness of this approach in most clinical settings, including emergency departments (EDs).

Although our understanding of the impact of performing nontargeted HIV screening in EDs has been improved over the past several years, we still have little understanding of specific individual-level characteristics that may influence the performance of this important preventive intervention. In fact, several studies have reported varying proportions of testing when the testing is performed in an ED environment and have highlighted differences in the proportions of patients who accept HIV testing and those who actually complete testing.^7–10 Perceived risk by the patient and the patient's medical acuity likely contribute to the relatively small proportion of patients who accept HIV testing, and several operational considerations likely prevent many of those who accept testing to actually complete testing. Other considerations that may importantly contribute to patient acceptance of nontargeted HIV screening include the ability to pay, specific demographic characteristics, and socioeconomic status.¹¹

The primary goal of this study was to estimate associations between payer status, race/ethnicity, and acceptance of nontargeted opt-out rapid HIV screening when performed in the ED. The secondary goal was to estimate associations between payer status, race/ethnicity, and completion of nontargeted opt-out rapid HIV screening in the ED.

METHODS

This was a planned secondary analysis of data collected from a prospective controlled clinical trial performed in the ED at Denver Health Medical Center in Denver, Colorado, between April 15, 2007, and April 15, 2009. The primary study used a quasi-experimental equivalent time-samples design to evaluate the effectiveness and efficiency of performing nontargeted opt-out rapid HIV screening in a large, urban ED compared with physician-directed diagnostic rapid HIV testing.¹² The 2 HIV testing methods were alternated sequentially every 4 months over the 2-year study period. For the purpose of this analysis, we included only those patients enrolled during the 12-month nontargeted opt-out rapid HIV screening phase of the trial because HIV screening was not performed during the diagnostic testing phase. HIV tests were provided to patients free of charge. Details of the design, implementation, and rationale as well as the primary results of the study have been previously reported.^9,12 This study was approved by our institutional review board.

Setting and Study Population

Denver Health Medical Center is a 477-bed urban teaching hospital with approximately 55 000 adult ED patient visits per year. Denver Health Medical Center is also a regional level-1 trauma center and a well-known model for the integration of a public hospital, community health center clinics, and a public health department.¹³ Approximately 70% of patients served by Denver Health Medical Center are racial or ethnic minorities and approximately 40% are uninsured.

During the nontargeted, opt-out, rapid HIV screening phase of the study, all patients aged 16 years and older and capable of providing consent for general emergency medical care were offered rapid HIV testing by registration personnel using an opt-out consent approach. Patients were excluded and thus not offered rapid HIV testing if they were

1. unable to provide consent as determined by registration or clinical staff (e.g., because of critical illness or altered mentation),

2. a prisoner or detainee,

3. a victim of sexual assault,

4. sought care as a result of an occupational exposure to HIV,

5. self-identified as being infected with HIV, or

6. left the ED before being placed in a treatment room.

Nontargeted Rapid HIV Screening

All patients who presented to the ED and who met criteria for inclusion were informed during the registration process that rapid HIV testing would be performed unless declined. An interpreter was used in instances where patients were non–English speaking. Patients were not aware of the study and, regardless of whether they consented to testing, received a 1-page informational sheet describing the epidemiology, transmission characteristics, and details of testing for HIV infection.

Registration staff used an indicator built into the electronic ED patient tracking system (Emergency Medical Services Information System [EMeSIS], Denver Health, Denver, CO) to indicate which patients agreed to HIV testing. This system was available to all ED staff and used additionally by nurses and health care technicians to identify such patients. No system was developed to electronically track patients who had been previously tested for HIV. For those patients who agreed to be tested, a blood sample was supposed to be obtained and sent to the hospital's laboratory for rapid HIV testing, which defined completed testing. Unfortunately, several reasons may have contributed to patient acceptance without completing testing (e.g., change in a patient's medical condition, discharge before laboratory testing, a patient deciding not to be tested after initially accepting testing).

For those patients who opted out, physicians had the additional opportunity to perform diagnostic testing (i.e., testing based on clinical signs or symptoms). Rapid HIV testing was performed by the hospital's laboratory, by using a sequential rapid HIV testing approach, and reported according to standard hospital's laboratory reporting procedures. All patients with preliminary positive results regardless of payer status had additional blood drawn in the ED for confirmatory Western blot testing and were linked into ongoing medical care by ED social workers.¹⁴

Data Collection

Patient characteristics were collected, including demographics (age, sex, and race/ethnicity), payer status, day of the week, and time of day. Registration personnel, using a standardized approach, collected these data and input them into an electronic hospital-based computer system during the registration process. Race/ethnicity was self-reported and classified according to 1 of the following categories: Black, American/Alaskan Native, Asian, Hispanic/Latino/Spanish, Native Hawaiian, other Pacific Islander, White, or unknown. Payer status was classified as commercial (i.e., private), Colorado Indigent Care Program (CICP; i.e., state-sponsored), Medicaid, Medicare, self, other, or missing. The CICP is not a health insurance program, but instead is a state-based program that distributes federal and state funds to partially compensate qualified health care providers for uncompensated costs associated with medical services rendered to those considered indigent. To qualify for CICP, applicants must (1) be a Colorado resident, (2) be a citizen of the United States, (3) have income and resources combined at or below 250% of the federal poverty level,^15–17 and (4) otherwise not be eligible for Medicaid. In effect, patients who qualify for CICP are, like self-pay patients, uninsured.

The primary outcome for this analysis was whether the patient agreed to (i.e., did not opt out of) HIV testing. Because not all patients who agreed to HIV testing actually completed testing, the secondary outcome was whether rapid HIV testing was completed.

Data Management and Statistical Analyses

We transferred data electronically or entered the data manually into a database (Microsoft Access 2003, Microsoft Corporation Inc, Redmond, WA) and transferred the data into SAS formats by using translational software (dfPower/DBMS Copy version 8, DataFlux Corporation, Cary, NC). We performed all statistical analyses with SAS version 9.2 (SAS Institute Inc, Cary, NC).

We expressed descriptive statistics for continuous variables as medians with interquartile ranges (IQRs) and proportions as percentages with 95% confidence intervals (CIs). We reported bivariate comparisons as absolute differences with 95% CIs. We used multivariable logistic regression with generalized estimating equations to model the associations between payer categories and whether patients accepted and completed rapid HIV testing. We made adjustments for age, sex, race/ethnicity, day of week, and time of day. We included day and time as covariates in the multivariable models to adjust for specific days of the week or times of the day as potential confounders for acceptance of HIV testing. Because a large number of repeat ED visits occurred during the study period, we employed a repeated measures approach to modeling using patient as the cluster. We used multiple imputation with SAS-callable IVEware version 0.2 (Survey Research Center, Institute for Social Research, University of Michigan, Ann Arbor) in instances where data were missing to minimize bias and preserve study power.¹⁸ We evaluated effect modification and included interaction terms in the final model if they contributed significantly to the model as defined by a P < .05. In addition, we used regression diagnostics and goodness-of-fit testing to assess the quality of each model. We did not calculate a priori sample size or perform adjustments for multiple comparisons. The unit of analysis was the patient unless stated otherwise, but because payer status may change between visits, these results are reported at the patient-visit level.

RESULTS

During the 12-month study period, 33 580 patients aged 16 years and older presented to the ED and were placed in a treatment room. Of these, 31 525 (92%) patients met criteria for inclusion and thus represent our study sample. A total of 44 765 unique ED visits were made by this cohort of patients. The median age of the patients was 37 (IQR = 26–50) years, 55% were male, and 40% were White, 37% Hispanic, 14% Black, 1% Asian, and 7% represented another or unknown race or ethnicity.

Of the 44 765 unique patient visits, 10 237 (23%) patients did not opt out of rapid HIV testing and 7656 (17%) completed rapid HIV testing. Of the 7656 total HIV tests performed, 561 (7%) were repeat tests, representing 456 unique patients. Of the 44 765 patient visits, 27% were self-pay, 23% were state-sponsored, 18% had Medicaid, 13% had commercial insurance, 12% had Medicare, and 8% had another, primarily governmental, or an unknown payer source.

Table 1 shows patient demographics and payer status of those who accepted, or did not opt out of, rapid HIV testing during the study period. Compared with the proportion of female patients who accepted HIV testing, a smaller proportion of male patients accepted HIV testing (difference = −3.8%; 95% CI = −2.8%, −4.7%). In addition, compared with the proportion of White patients who accepted HIV testing, larger proportions of Black patients (difference = 5.4%; 95% CI = 3.9%, 7.0%) and Hispanic patients (difference = 4.6%; 95% CI = 3.5%, 5.7%) accepted HIV testing.

TABLE 1—

Patient Demographics and Payer Status Stratified by Those Who Accepted Nontargeted Opt-Out Rapid HIV Screening: Emergency Department, Denver Health Medical Center, 2007–2009

Characteristic	Did Not Opt Out of Rapid HIV Testing, No. (%) or Median (IQR)	Opted Out of Rapid HIV Testing, No. (%) or Median (IQR)	Total No.
Patient level	7962	23 563	31 525
Age, y	36 (26–48)	38 (26–51)	31 525
Sex
Male	4049 (24)	13 150 (76)	17 199
Female	3912 (27)	10 407 (73)	14 319
Missing	1 (14)	6 (86)	7
Race/ethnicity
Asian	67 (15)	376 (85)	443
Black	1270 (29)	3151 (71)	4421
Hispanic	3273 (28)	8485 (72)	11 758
Othera	109 (26)	310 (74)	419
White	2957 (23)	9759 (77)	12 716
Missing	286 (16)	1482 (84)	1768
Patient visit level	10 237	34 528	44 765
Payer status
Commercial	1092 (19)	4538 (81)	5630
Medicaid	1791 (23)	6100 (77)	7891
Medicare	957 (18)	4471 (82)	5428
Self-pay	3287 (27)	8887 (73)	12 174
Stateb	2721 (27)	7442 (73)	10 163
Otherc	219 (10)	1909 (90)	2128
Missing	170 (13)	1181 (87)	1351

Note. IQR = interquartile range.

^aDefined as American or Alaskan Native, Native Hawaiian, or Non-Hawaiian Pacific Islander.

^bDefined as the Colorado Indigent Care Program.

^cDefined as payment provided by the City Attorney (e.g., victims of crime), Department of Safety, worker's compensation, or other governmental sources.

Compared with the proportion of patient visits that represented a commercial payer source and who accepted HIV testing, a larger proportion of visits representing self-pay (difference = 7.6%; 95% CI = 6.3%, 8.9%), state sponsorship (difference = 7.3%; 95% CI = 6.0%, 8.7%), and Medicaid (difference = 3.3%; 95% CI = 1.9%, 4.7%) accepted HIV testing. We operationally defined uninsured patients as those classified as self-pay or state-sponsored. As such, of the proportion of patient visits that were uninsured, 27% accepted HIV testing; by comparison, among patient visits that were insured, only 19% accepted HIV testing (difference = 7.6%; 95% CI = 6.9%, 8.4%).

Table 2 shows patient demographics and payer status for those who did and did not complete rapid HIV testing during the study period. Similar to those who accepted HIV testing, compared with the proportion of female patients who completed HIV testing, a smaller proportion of male patients completed testing (difference = −3.1; 95% CI = −2.3%, −4.1%). In addition, compared with the proportion of White patients who completed HIV testing, larger proportions of Black patients (difference = 4.7%; 95% CI = 3.3%, 6.1%) and Hispanic patients (difference = 3.9%; 95% CI = 2.9%, 4.9%) completed HIV testing.

TABLE 2—

Patient Demographics and Payer Status Stratified by Those Who Completed Rapid HIV Testing When Offered Using a Nontargeted Opt-Out Rapid HIV Screening Approach: Emergency Department, Denver Health Medical Center, 2007–2009

Characteristic	Completed Rapid HIV Testing, No. (%) or Median (IQR)	Did Not Complete Rapid HIV Testing, No. (%) or Median (IQR)	Total No.
Patient level	5964	25 561	31 525
Age, y	37 (26–48)	37 (26–51)	31 525
Sex
Male	3004 (18)	14 195 (82)	17 199
Female	2959 (21)	11 360 (79)	14 319
Missing	1 (14)	6 (86)	7
Race/ethnicity
Asian	48 (11)	395 (89)	443
Black	968 (22)	3453 (78)	4421
Hispanic	2477 (21)	9281 (79)	11 758
Othera	83 (20)	336 (80)	419
White	2182 (17)	10 534 (83)	12 716
Missing	206 (12)	1562 (88)	1768
Patient visit level	7656	37 109	44 765
Payer status
Commercial	781 (14)	4849 (86)	5630
Medicaid	1360 (17)	6531 (83)	7891
Medicare	718 (13)	4710 (87)	5428
Self-pay	2424 (20)	9750 (80)	12 174
Stateb	2115 (21)	8048 (79)	10 163
Otherc	146 (7)	1982 (93)	2128
Missing	112 (8)	1239 (92)	1351

Note. IQR = interquartile range.

^aDefined as American or Alaskan Native, Native Hawaiian, or Non-Hawaiian Pacific Islander.

^bDefined as the Colorado Indigent Care Program.

^cDefined as payment provided by the City Attorney (e.g., victims of crime), Department of Safety, worker's compensation, or other governmental sources.

Compared with the proportion of patient visits that represented a commercial payer source and who completed HIV testing, a larger proportion of visits representing self-pay (difference = 6.0%; 95% CI = 4.9%, 7.2%), state sponsorship (difference = 6.9%; 95% CI = 5.7%, 8.1%), and Medicaid (difference = 3.4%; 95% CI = 2.1%, 4.6%) completed HIV testing.

Table 3 shows patient demographics by payer status. Of the 18 003 visits by White patients, 27% were self-pay, 21% had CICP, 19% had commercial insurance, 14% had Medicaid, and 13% had Medicare. Of the 16 282 visits by Hispanic patients, 30% were self-pay, 27% had CICP, 21% had Medicaid, 11% had Medicare, and only 8% had commercial insurance. Finally, of the 6495 visits by Black patients, 25% were self-pay, 23% had CICP, 23% had Medicaid, 17% had Medicare, and only 7% had commercial insurance. In addition, of the 22 337 uninsured patient visits (defined as self-pay or CICP), 41% were by Hispanic patients, 39% by White patients, and 14% by Black patients. Of the 5630 commercial patient visits, 61% were by White patients. Compared with the proportion of visits by White patients who were uninsured, a similar proportion of visits by Black patients were uninsured (difference = 0.6%; 95% CI = −0.9%, 2.0%) and a larger proportion of Hispanic patients were uninsured (difference = 7.8%; 95% CI = 6.8%, 8.9%).

TABLE 3—

Patient Demographics by Payer Status, Emergency Department, Denver Health Medical Center, 2007–2009

Characteristic	Commercial, No. (%) or Median (IQR)	Medicaid, No. (%) or Median (IQR)	Medicare, No. (%) or Median (IQR)	State,a No. (%) or Median (IQR)	Self-Pay, No. (%) or Median (IQR)	Other,b No. (%) or Median (IQR)	Missing, No. (%) or Median (IQR)	Total, No. (%) or Median (IQR)
Total	5630 (13)	7891 (18)	5428 (12)	10 163 (23)	12 174 (27)	2128 (5)	1351 (2)	44 765
Agec	32 (23–46)	40 (25–52)	60 (49–71)	42 (32–50)	33 (24–44)	37 (28–47)	38 (25–51)	40 (27–51)
Sex
Male	3009 (12)	3186 (13)	2942 (12)	5753 (23)	7693 (31)	1564 (6)	802 (3)	24 949
Female	2621 (13)	4704 (24)	2483 (13)	4400 (22)	4477 (23)	564 (3)	549 (2)	19 798
Missing	0 (0)	1 (6)	3 (17)	10 (56)	4 (21)	0 (0)	0 (0)	18
Race/ethnicity
Asian	102 (19)	134 (26)	54 (10)	88 (17)	126 (24)	20 (4)	0 (0)	524
Black	472 (7)	1488 (23)	1083 (17)	1490 (23)	1611 (25)	347 (5)	4 (0)	6495
Hispanic	1330 (8)	3338 (21)	1839 (11)	4326 (27)	4810 (30)	630 (3)	9 (0)	16 282
White	3446 (19)	2552 (14)	2279 (13)	3772 (21)	4924 (27)	1021 (6)	9 (0)	18 003
Otherd	69 (9)	156 (20)	66 (8)	203 (26)	268 (34)	16 (21)	0 (0)	778
Missing	211 (8)	223 (8)	107 (4)	264 (10)	435 (16)	94 (4)	1329 (50)	2663

Note. IQR = interquartile range. Patient visits constitute the unit of analysis for data reported in this table.

^aDefined as the Colorado Indigent Care Program.

^bDefined as payment provided by the City Attorney (e.g., victims of crime), Department of Safety, worker's compensation, or other governmental sources.

^cReported as median with interquartile range in parentheses.

^dDefined as American or Alaskan Native, Native Hawaiian, or Non-Hawaiian Pacific Islander.

Table 4 shows the results from 2 multivariable logistic regression models to estimate associations between payer status and whether patients agreed to or completed HIV testing with adjustment for age, gender, race/ethnicity, and day or time of the week. Patients who were self-pay, state-sponsored, or had Medicaid had increased odds of accepting or completing HIV testing relative to those with commercial insurance. Also, Black and Hispanic patients had increased odds of accepting or completing HIV testing relative to White patients. In contrast, Asian patients had decreased odds of accepting or completing HIV testing relative to White patients.

TABLE 4—

Repeated Measures Multivariable Logistic Regression to Estimate Associations Between Payer Status and Patient Demographics and Agreement to or Completion of HIV Testing in the Emergency Department, Denver Health Medical Center, 2007–2009

Characteristic	Did Not Opt Out of Rapid HIV Testing, OR (95% CI)	Completed Rapid HIV Testing, OR (95% CI)
Payer status
Commercial (Ref)	1.00	1.00
Medicaid	1.24 (1.14, 1.34)	1.32 (1.20, 1.45)
Medicare	1.07 (0.96, 1.18)	1.12 (1.00, 1.26)
Statea	1.64 (1.52, 1.77)	1.77 (1.62, 1.93)
Self-pay	1.63 (1.51, 1.75)	1.65 (1.52, 1.80)
Otherb	0.27 (0.23, 0.31)	0.24 (0.20, 0.29)
Age	0.99 (0.99, 1.00)	0.99 (0.99, 1.00)
Sex
Male	0.82 (0.78, 0.86)	0.82 (0.78, 0.86)
Female (Ref)	1.00	1.00
Race/ethnicity
Asian	0.64 (0.50, 0.82)	0.61 (0.46, 0.81)
Black	1.29 (1.21, 1.38)	1.30 (1.21, 1.40)
Hispanic	1.17 (1.11, 1.23)	1.18 (1.11, 1.24)
Otherc	0.91 (0.76, 1.08)	0.89 (0.73, 1.09)
White (Ref)	1.00	1.00

Notes. CI = confidence interval; OR = odds ratio. Age was modeled as a continuous variable and, in addition to the variables listed in the table, both models included day of the week and time of day as covariates. The Hosmer-Lemeshow goodness-of-fit P values for the 2 models were 0.84 and 0.81, respectively.

^aDefined as the Colorado Indigent Care Program.

^bDefined as payment provided by the City Attorney (e.g., victims of crime), Department of Safety, worker's compensation, or other governmental sources.

^cDefined as American or Alaskan Native, Native Hawaiian, or Non-Hawaiian Pacific Islander.

DISCUSSION

Our study represents the largest evaluation to date of payer status and race/ethnicity on acceptance of nontargeted, opt-out HIV screening in an ED setting, and demonstrates that a large proportion of patients who receive care in EDs are uninsured and that a significant proportion of uninsured patients accept HIV testing in this clinical setting. Our results also show that Hispanic patients are more likely to be uninsured or subsidized through government programs than are White patients, yet they are more likely to consent to and complete free rapid HIV testing when offered as part of a nontargeted opt-out screening program. In addition, although the proportion of uninsured Black patients is similar to that of White patients, Black patients were also more likely to consent to and complete free rapid HIV testing.

The HIV epidemic in the United States remains an important public health priority and HIV testing continues to be a major prevention focus. Populations considered most at risk for acquiring HIV infection include, in part, racial/ethnic minorities and those who are socioeconomically disadvantaged.¹⁹ Currently, approximately 46 million individuals remain uninsured in the United States and most of these individuals lack routine primary care.²⁰ Such patients commonly use EDs as their sole source of medical care, resulting in a significant proportion of the estimated 120 million ED visits annually in the United States.²¹

Since 2006, a substantial push to incorporate widespread “routine” opt-out HIV screening has been led by the CDC.⁶ Much of the focus has included performing nontargeted HIV screening in EDs. Although we acknowledge the theoretical benefit of widespread HIV screening on mitigating the HIV epidemic in the United States, we also recognize significant financial constraints related to such an approach. What has remained largely unanswered is how such large-scale preventive services will be subsidized.

Our results demonstrated a significantly larger association between uninsured patients and acceptance of HIV testing compared with insured patients. This raises a critical question related to how screening efforts may be initiated, expanded, or sustained when financial resources are scarce. Patients without health insurance will likely find the charges associated with rapid HIV testing in the ED prohibitively high. Because of the large number of low-risk patients offered testing with a nontargeted screening approach, we believe that a more effective and efficient testing method is warranted (i.e., one that focuses scarce resources on patients most at risk). In addition, hospital and ED administrators are unlikely to support nontargeted screening initiatives if they are not cost-neutral, at best, and current methods for funding such screening initiatives rely heavily on state and federal funding.

Although “routine” HIV screening has been shown to be generally cost-effective from a societal perspective,^22,23 it remains unclear what effect, if any, the passage of the Affordable Care Act²⁴ will have on specific health interventions, including HIV testing. This bill requires most new individual health insurance policies and employer-sponsored health plans to offer, free of charge, “HIV screening for those at risk of the disease.” Unfortunately, this act does not incorporate HIV testing into existing policies and falls short of subsidizing HIV testing for those not identified as being at risk, the precise model proposed by the CDC. It therefore still remains unclear as to how and to what extent nontargeted HIV screening will be subsidized if this approach to testing continues to be endorsed. Furthermore, those most likely to accept nontargeted HIV screening are those most likely to not to be able to pay for it, thus suggesting the need to identify the most cost-effective method of performing HIV testing in this clinical setting.

Black and Hispanic patients disproportionately represent the highest number of new HIV infections in the United States. Despite making up only 13% of the total population, Black patients account for 41% of all new HIV infections and 48% of all patients known to be infected with HIV.²⁵ In addition, Hispanic patients, despite representing only 13% of the total population, account for 19% of new infections and 17% of all patients infected with HIV.^25,26 Thus, our findings that being Black or Hispanic is positively associated with accepting and completing rapid HIV testing in an ED setting is encouraging and partially supports the utility of providing this service in ED settings.

In addition, research has shown that socioeconomic status is an important determinant of HIV risk.²⁷ Patients living in poverty often have poorer health status and increased disparities in medical care experiences.^28–30 Patients living in poverty are also often unable to pay for health care services, which may result in not seeking proper care when needed to prevent acquisition of HIV infection or its long-term complications.²⁷ In 2006, 24% of Black individuals and 23% of Hispanic individuals lived in poverty. In our study, 30% of visits by Hispanic and 25% by Black patients were self-pay and an additional 59% of visits by Hispanic and 63% by Black patients were Medicare, Medicaid, or subsidized through our indigent state-sponsored program. Although it is encouraging that such a large majority of these patients were more likely to accept and complete rapid HIV testing than those with commercial insurance, it remains unclear how such testing will be subsidized.

There was a relatively low proportion of acceptance of HIV testing in our study, although other major studies have reported similar acceptance results.^2,10 Furthermore, though approximately 2000 patients did not complete testing after accepting it, differences in the proportions of patients who accepted or completed HIV testing among patient characteristics (i.e., demographics and payer status) did not appear to differ appreciably. Instead, we believe the difference between those who accepted testing and those who completed testing largely represents operational constraints of performing this large preventive intervention in a high-volume urban ED. In addition, the adjusted associations between payer status and race/ethnicity, and acceptance and completion of HIV testing remained similar.

Limitations

Our study should be interpreted in the context of several potential limitations. The main limitation of this study is that it was a secondary analysis of a data set that was not specifically designed to determine which variables have an impact on the decision to accept HIV testing. Another limitation includes possible misclassification of payer or race/ethnicity variables. We believe this possibility was minimized by the large number of consecutive patients included in the study as well as the use of trained registration personnel who obtained these data by using a standardized approach.

Other characteristics (e.g., employment, income, housing) were also not collected and thus were not included in the multivariable models. It is possible that payer status or race/ethnicity served as a confounder for these other social determinants. Finally, this study was performed at a single institution and the results, therefore, may not reflect results obtained from other institutions or communities.

Conclusions

A large proportion of patients who present to the ED are uninsured or subsidized through government programs and are more likely than patients with commercial insurance to consent to and complete free rapid HIV testing when offered as part of a nontargeted opt-out screening program. Although efforts to improve payment for nontargeted HIV screening are ongoing, including improving commercial payment, additional efforts to fund HIV testing or focus scarce resources are required.