HIV Testing by Transgender Status at Centers for Disease Control and Prevention–Funded Sites in the United States, Puerto Rico, and US Virgin Islands, 2009–2011

Abstract

Objectives. We examined HIV testing services, seropositivity, and the characteristics associated with newly identified, confirmed HIV-positive tests among transgender individuals.

Methods. We analyzed data (2009–2011) using bivariate and multivariable logistic regression to examine the relationships between HIV positivity and sociodemographic and risk characteristics among male-to-female transgender individuals.

Results. Most of the testing was conducted in females (51.1%), followed by males (48.7%) and transgender individuals (0.17%). Tests in male-to-female transgender individuals had the highest, newly identified confirmed HIV positivity (2.7%), followed by males (0.9%), female-to-male transgender individuals (0.5%), and females (0.2%). The associated characteristics with an HIV-positive test among male-to-female transgender individuals included ages 20 to 29 and 40 to 49 years (adjusted odds ratio [AOR] = 2.8; 95% confidence interval [CI] = 1.4, 5.6 and AOR = 2.8; 95% CI = 1.3, 5.9, respectively), African American (AOR = 4.6; 95% CI = 2.7, 7.9) or Hispanic/Latino (AOR = 2.6; 95% CI = 1.5, 4.5) race/ethnicity, and reporting sex without condom within the past year (AOR = 1.9; 95% CI = 1.3, 2.6), sex with an HIV-positive person (AOR = 1.5; 95% CI = 1.1, 2.0), or injection drug use (AOR = 2.0; 95% CI = 1.3, 3.0).

Conclusions. High levels of HIV positivity among transgender individuals, particularly male-to-female transgender individuals, underscore the necessity for targeted HIV prevention services that are responsive to the needs of this population.

The term “transgender” refers to a diverse group of people whose gender identity or expression does not conform to gender norms associated with their assigned gender at birth.¹ To date, the size of the transgender population in the United States,² and the national surveillance data with regard to the incidence and prevalence of HIV/AIDS for the transgender population, are unknown.³ Nonetheless, many studies suggest that the transgender population is disproportionately affected by HIV/AIDS. More than 1.1 million people are living with HIV in the United States, at an overall prevalence of 446.4 per 100 000 population.⁴ HIV prevalence among US male-to-female transgender individuals averaged 22% and 28%, respectively, across 2 meta-analyses,^3,5 and 2 studies of male-to-female transgender individuals found a prevalence of 39%.⁶ Self-reported HIV prevalence among male-to-female transgender individuals may be lower than rates of laboratory-confirmed infection; 1 meta-analysis found a self-reported prevalence of 12%,³ suggesting that many transgender individuals are unaware of their HIV status. Some subgroups may be particularly affected by HIV; 1 meta-analysis found a higher prevalence among male-to-female transgender sex workers (27%) than those not engaged in sex work (15%).⁷ The prevalence of HIV infection is higher among African American male-to-female transgender individuals than other racial/ethnic groups.^3,6,8,9 Lower levels were found among female-to-male transgender individuals, with self-report and laboratory-confirmed infection ranging between 0% and 3%.^3,10

Among male-to-female transgender individuals, HIV is associated with multiple behavioral risk factors.^3,11–13 One meta-analysis found high levels of reported risk behaviors among male-to-female transgender individuals, including sex with casual partners (48%), unprotected receptive anal intercourse (URAI; 44%), sex work (42%), sex while drunk or high (39%), and multiple sex partners (32%). Injection of street drugs (12%) was less commonly reported than that of hormones or silicone injections (27% and 25%, respectively), and needle sharing for street drugs, or hormones or silicone (9% and 6%, respectively) was found at lower levels.³

HIV risk behaviors among female-to-male transgender individuals are less studied. Reported behaviors include URAI (27%–29%),^9,10 unprotected vaginal intercourse (25%–63%),^9,10,14 sex work (19%–44%),^9,10,14 and injection of hormones with (54%)¹⁰ or without medical supervision (17%).⁹ Injection drug use (IDU) in female-to-male transgender individuals ranged from 0% to 21% across multiple studies.^9,10,15–17 Three studies found a reported prevalence of needle sharing among female-to-male transgender individuals between 0% and 5%.^9,16,18 Drug and alcohol use during sex was also reported in female-to-male transgender individuals.^14,18

Stigma and discrimination contribute to risk behaviors among transgender individuals; 1 study found nearly 40% of transgender respondents who experienced lifetime economic discrimination and nearly 60% who experienced lifetime violence or harassment.¹⁹ Stigma and discrimination affect transgender individuals’ ability to secure stable work and housing, which often results in financial vulnerability and engagement in commercial sex work.^3,10,20–24 Many transgender individuals experience discrimination in health care settings,^20,25 reporting that they do not seek,¹¹ postpone,²⁶ or are refused care,²⁶ or do not have access to needed services.²⁰ Transgender individuals may inject hormones or silicone without medical supervision, potentially increasing their risk for HIV infection and transmission.^3,10,27–29 Mental health issues, including depression and anxiety,^21,30,31 have been reported, and may be linked to high-risk sexual behavior and HIV.^22,32–34 Other factors that may contribute to risk include lack of knowledge about HIV,^2,11,22 and intimate partner violence^35,36 or gender abuse.³⁴

Because of these issues, complex HIV interventions for transgender individuals are critical.¹² Several recent interventions designed for male-to-female transgender individuals worked to address risk behavior^23,37,38 and contextual factors, such as financial and emotional hardships.³⁹ Culturally relevant interventions and health care services are needed.⁴⁰

HIV testing and diagnosis are critical steps along the continuum of HIV care, and can serve to facilitate linkage to and retention in care and treatment programs, and ultimately, viral suppression. However, little is known about the transgender population’s testing patterns. Large-scale studies on transgender testing that examine rates of HIV infection and factors associated with infection are not available. We described the characteristics of HIV testing events conducted among transgender populations using recent (2009–2011) data collected from HIV prevention programs funded by the Centers for Disease Control and Prevention (CDC) to support the development of targeted transgender HIV prevention and testing programs. Specifically, we described characteristics of all testing events and newly identified confirmed HIV-positive testing events that reported current gender as transgender in comparison with nontransgender testing events, and we also described the characteristics that were significantly associated with newly identified confirmed HIV-positive testing events among male-to-female transgender individuals.

METHODS

For this analysis we used data from 59 CDC-funded state and local health departments and community-based organizations that conducted HIV testing at health care and non–health care sites in 2009, 2010, and 2011.^41–44 Information about demographic characteristics and behavioral risk factors are collected about the persons tested, their current and previous test results, receipt of test results, and referrals. De-identified information on the tests is then reported by health departments to the CDC National HIV Prevention Program Monitoring and Evaluation System.

Our unit of analysis was an HIV testing event. An HIV testing event was a sequence of 1 or more HIV tests conducted with a client to determine their HIV status. During a testing event, a client could be tested once (e.g., 1 conventional test) or multiple times (e.g., 1 rapid test followed by 1 conventional test to confirm a preliminary HIV-positive test result). For these testing data, we could not link the results of repeat testing events for the same person if, for example, a person had more than 1 testing event that was represented in the 2009 to 2011 data. Thus, we considered these data to be test-level data rather than client-level data. Health departments had to have reported valid test-level data to be included in our analysis. An HIV test was considered invalid if all of the following variables had missing data: test election, test technology, specimen type, test result, and results received. Of the 9 473 668 testing events reported in 2009 to 2011, 9 400 761 were considered valid HIV testing events. Invalid HIV testing events were more likely to be associated with testing events for Whites, males, individuals from the Northeast and Midwestern regions, and tests conducted in health care settings. Records with a missing or invalid gender variable (75 382), or clients who were younger than 13 years (21 763) were excluded from study. Based on these criteria, 9 303 616 testing events were included in our analysis (Table 1).

TABLE 1—

Number of HIV Testing Events and Newly Confirmed HIV Testing Events by Select Characteristics of Persons Tested and Gender: United States, 2009–2011

	Malea		Femalea		MTF Transgendera		FTM Transgendera		All Transgendera		Totalb
Characteristics	No. (%)	Positivity, %	No. (%)	Positivity, %	No. (%)	Positivity, %	No. (%)	Positivity, %	No. (%)	Positivity, %	No. (%)
Age at test, y
13–19	376 668 (8.3)	0.4	625 834 (13.2)	0.1	1 032 (7.8)	1.2	188 (8.0)	0.0	1 220 (7.9)	1.0	1 003 722 (10.8)
20–29	1 749 073 (38.6)	0.9	2 076 629 (43.7)	0.1	6 129 (46.6)	3.0	1 262 (53.4)	0.3	7 391 (47.6)	2.5	3 833 093 (41.2)
30–39	967 863 (21.3)	1.0	980 384 (20.6)	0.3	3 176 (24.1)	3.1	556 (23.5)	0.5	3 732 (24.0)	2.7	1 951 979 (21.0)
40–49	756 004 (16.7)	1.1	602 899 (12.7)	0.5	1 690 (12.8)	2.8	216 (9.1)	0.9	1 906 (12.3)	2.6	1 360 809 (14.6)
≥ 50	648 610 (14.3)	0.8	433 260 (9.1)	0.5	998 (7.6)	1.5	116 (4.9)	1.7	1 114 (7.2)	1.5	1 082 984 (11.6)
Race/ethnicity
African American	1 919 872 (42.3)	1.0	2 038 248 (42.9)	0.4	4 573 (34.8)	4.2	563 (23.8)	0.7	5 136 (33.1)	3.9	3 963 256 (42.6)
Hispanic/Latino	882 857 (19.5)	0.9	971 778 (20.4)	0.2	4 397 (33.4)	2.5	398 (16.8)	0.8	4 795 (30.9)	2.3	1 859 430 (20.0)
White	1 294 078 (28.5)	0.7	1 189 313 (25.0)	0.1	2 385 (18.1)	1.0	1 102 (46.6)	0.4	3 487 (22.5)	0.8	2 486 878 (26.7)
Other	286 241 (6.3)	0.8	264 323 (5.6)	0.2	1 791 (13.6)	1.8	300 (12.7)	0.0	2 091 (13.5)	1.5	552 655 (5.9)
Region
Northeast	1 101 975 (24.3)	0.7	984 447 (20.7)	0.3	4 311 (32.8)	2.5	983 (41.6)	0.5	5 294 (34.1)	2.2	2 091 716 (22.5)
Midwest	508 563 (11.2)	0.6	433 147 (9.1)	0.1	1 284 (9.8)	2.4	173 (7.3)	0.0	1 457 (9.4)	2.1	943 167 (10.1)
South	2 350 933 (51.8)	1.0	2 990 909 (62.9)	0.3	3 627 (27.6)	3.4	508 (21.5)	0.4	4 135 (26.6)	3.0	5 345 977 (57.5)
West	521 771 (11.5)	1.0	287 916 (6.1)	0.2	3 734 (28.4)	2.4	691 (29.2)	0.6	4 425 (28.5)	2.1	814 112 (8.8)
US dependent areas	51 184 (1.1)	1.5	57 253 (1.2)	0.5	198 (1.5)	3.0	9 (0.4)	0.0	207 (1.3)	2.9	108 644 (1.2)
Test site type
Health care facilities	2 582 740 (57.0)	0.9	3 543 013 (74.5)	0.2	4 976 (37.8)	2.4	1 020 (43.1)	0.4	5 996 (38.6)	2.1	6 131 749 (65.9)
Non–health care facilities	1 196 434 (26.4)	1.1	844 628 (17.8)	0.4	7 414 (56.4)	2.8	1 211 (51.2)	0.6	8 625 (55.6)	2.5	2 049 687 (22.0)
Correctional facilities	599 755 (13.2)	0.4	160 291 (3.4)	0.4	394 (3.0)	5.1	32 (1.4)	0.0	426 (2.7)	4.7	760 472 (8.2)
Other facilities	88 818 (2.0)	0.9	106 486 (2.2)	0.1	295 (2.2)	2.7	91 (3.8)	0.0	386 (2.5)	2.1	195 690 (2.1)
Receipt of test result: yes	3 252 911 (85.0)c		2 825 392 (76.1)c		12 117 (94.8)c		2 170 (95.4)c		14 287 (94.9)c		6 092 590 (80.6)c
Total	4 534 426 (48.7)	0.9	4 753 672 (51.1)	0.2	13 154 (0.1)	2.7	2 364 (0.03)	0.5	15 518 (0.2)	2.4	9 303 616 (100.0)

Note. FTM = female-to-male; MTF = male-to-female.

^aBecause of missing or invalid data, the values in each column may not sum to the column total.

^bExcludes HIV testing events with missing and invalid values (75 382) for the variable of gender.

^cExcludes HIV testing events with missing and invalid values for the variable of receipt of test result.

Variables Analyzed

In 2008, health departments were required to submit new HIV testing variables to the CDC to inform programming and HIV testing initiatives. The gender variable was revised to include male-to-female transgender and female-to-male transgender as response options, in addition to male and female genders. Gender is defined as the self-reported current gender identity of the client, which might include one’s social status, self-identification, legal status, or biology. We recognized that various terms (e.g., transgender man, transgender woman) might be used in the literature or preferred by the transgender community, and that the 2 categories we reported herein might not fully represent the complexities of transgender identities; however, in this article, male-to-female transgender and female-to-male transgender are used because these terms were used in CDC data collection requirements. Additional demographic characteristic information collected included age at the time of testing, race/ethnicity, and geographic location of the HIV test.

We categorized a testing event as a newly identified confirmed HIV-positive result if a confirmed HIV-positive test result was associated with a client who did not self-report having previously tested as HIV positive. We grouped the testing site type where HIV testing was conducted into the following facility categories: (1) health care (includes inpatient and outpatient facilities, and emergency rooms), (2) non–health care (includes HIV counseling and testing sites, and community settings), (3) correctional (includes any correctional facility with or without a health care unit), and (4) other (includes blood banks or plasma centers and other facilities).

The receipt of an HIV test result was a calculated variable that indicated whether clients received their HIV test result for at least 1 HIV test in the testing event. We collected risk behavior information from clients at the time of testing, with a 12-month recall period. Clients could report none, 1, or multiple risk behaviors, or refuse to answer these questions. Health departments were required to report risk behavior information only from clients tested in non–health care settings and individuals who tested positive in health care settings; however, many health departments, regardless of this requirement, reported client risk behavior information for both settings.

Analysis

We calculated descriptive statistics on all testing events for each gender group to examine the distribution of sociodemographic and risk behavior characteristics, testing site type, and geographic region. We conducted bivariate analyses to examine associations between a newly identified confirmed HIV-positive testing event among male-to-female transgender individuals and sociodemographic and risk behavior characteristics, testing site type, and geographic region. We did not conduct these analyses with data from female-to-male transgender individuals because of small cell sizes on demographic and risk characteristics. We estimated multiple logistic regression models to determine the association between male-to-female transgender characteristics, testing site, and geographic region and the likelihood of having a newly identified confirmed HIV-positive testing event. We included all predictor variables that showed significant bivariate association (at P < .05 level) in the model. We used odds ratios (ORs) and 95% confidence intervals (CIs) to evaluate the significance of the associations between the predictor variables and the likelihood of HIV positivity among testing events from male-to-female transgender individuals. We evaluated the overall fit of the multivariable model using the Hosmer and Lemeshow goodness-of-fit test.⁴⁵

RESULTS

In 2009 to 2011, a total of 9 303 616 testing events were reported to the CDC: 4 534 426 (48.7%) among males, 4 753 672 (51.1%) among females, and 15 518 (0.17%) among transgender individuals (Table 1). Of the transgender testing events, 0.03% were among female-to-male transgender individuals and 0.14% were among male-to-female transgender individuals. Seventy-five percent of the total testing events reported risk behavior information.

Overall Testing Patterns and Positivity by Gender

Of all reported testing events, those among transgender individuals (15 518) had the highest newly identified confirmed HIV-positivity across genders (2.4%), followed by male (0.9%) and female (0.2%) individuals (Table 1). The majority of all testing events, regardless of gender, were conducted among those ages 20 to 39 years in the Northeast and South regions. The majority of male and female testing events were conducted among African Americans (1 919 872, 42.3% and 2 038 248, 42.9%, respectively) and Whites (1 294 078, 28.5% and 1 189 313, 25.0% respectively), and at health care facilities (2 582 740, 57.0% and 3 543 013, 74.5%, respectively). The majority of transgender testing events were conducted among African Americans (5136, 33.1%) and Hispanic/Latinos (4795, 30.9%), and at non–health care facilities (8625, 55.6%). The percentage of all testing events that were followed up with receipt of HIV test results was highest among transgender individuals (14 287, 94.9%), followed by male individuals (3 252 911, 85.0%), and lowest among female individuals (2 825 392, 76.1%; Table 1).

Among 15 518 transgender testing events, 13 154 (84.8%) of the events were among male-to-female transgender individuals, and the remaining were among female-to-male transgender individuals (2364, 15.2%; Table 1). Among male-to-female transgender individuals, the highest percentage of all HIV testing events conducted were among African Americans (4573, 34.8%), followed by Hispanics/Latinos (4397, 33.4%) and Whites (2385, 18.1%); conversely, among female-to-male transgender individuals, the highest percentage of testing events were conducted among Whites (1102, 46.6%), followed by African Americans (563, 23.8%) and Hispanics/Latinos (398, 16.8%). Male-to-female transgender individuals reported the highest percentage of testing events in the Northeast (4311, 32.8%), followed by the West (3734, 28.4%) and the South (3627, 27.6%); similarly, female-to-male transgender testing events occurred primarily in the Northeast (983, 41.6%), followed by the West (691, 29.2%) and the South (508, 21.5%). Among all transgender testing events, male-to-female transgender individuals had the highest overall newly identified confirmed HIV-positivity (2.7%). Female-to-male transgender individuals with newly identified confirmed HIV-positivity was 0.5% (Table 1).

Risk Behaviors in Past 12 Months by Gender

Risk behavior data analysis is based on the 6 978 098 testing events that reported risk behavior information from 2009 to 2011 (Table 2). The most frequently reported risk behaviors for males, females, and transgender individuals included sex without a condom (60.4%, 59.8%, and 60.0%, respectively), sex with a person of unknown HIV status (14.1%, 11.0%, and 18.9%, respectively), and sex while intoxicated or high on drugs (12.5%, 6.2%, and 16.0%, respectively). Transgender individuals reported the following risk behaviors at higher frequency than males or females: exchange of sex for drugs, money, or something they used (11.0%); sex with an anonymous partner (9.7%); and sex with a person who was HIV-positive (9.0%; Table 2).

TABLE 2—

Number of HIV Testing Events and Newly Confirmed HIV Testing Events by Risk Behaviors of Persons Tested and Gender: United States, 2009–2011

	Male		Female		MTF Transgender		FTM Transgender		All Transgender		Totala
Characteristic	No.b (%)	Positivity, %	No.b (%)	Positivity, %	No.b (%)	Positivity, %	No.b (%)	Positivity, %	No.b (%)	Positivity, %	No.b (%)
Sex without using condom
Yes	2 065 360 (60.4)	0.9	2 118 517 (59.8)	0.3	7 241 (61.5)	3.1	1 047 (51.1)	0.4	8 288 (60.0)	2.8	4 192 165 (60.1)
No	752 358 (22.0)	0.8	674 584 (19.0)	0.3	2 706 (23.0)	1.7	567 (27.7)	0.2	3 273 (23.7)	1.4	1 430 215 (20.5)
Sex with IDU
Yes	135 238 (4.0)	1.0	119 051 (3.4)	0.4	628 (5.3)	3.7	89 (4.3)	2.2	717 (5.2)	3.5	255 006 (3.7)
No	2 842 392 (83.1)	0.9	3 090 738 (87.2)	0.2	9 133 (77.6)	2.8	1 591 (77.7)	0.3	10 724 (77.6)	2.4	5 943 854 (85.2)
Sex with person who is HIV positive
Yes	126 087 (3.7)	4.7	44 311 (1.3)	2.9	1 084 (9.2)	5.6	157 (7.7)	0.6	1 241 (9.0)	5.0	171 639 (2.5)
No	2 877 037 (84.1)	0.8	3 177 145 (89.6)	0.2	8 961 (76.1)	2.6	1 623 (79.3)	0.4	10 584 (76.6)	2.2	6 064 766 (86.9)
IDU in past 12 mo
Yes	166 418 (4.9)	0.9	101 874 (2.9)	0.5	670 (5.7)	4.6	128 (6.3)	0.8	798 (5.8)	4.0	269 090 (3.9)
No	2 909 471 (85.1)	0.9	3 178 558 (89.7)	0.2	9 899 (84.1)	2.8	1 745 (85.2)	0.4	11 644 (84.3)	2.4	6 099 673 (87.4)
Exchange sex for drugs/money/other
Yes	60 780 (1.8)	1.4	66 221 (1.9)	0.8	1 450 (12.3)	3.2	67 (3.3)	1.5	1 517 (11.0)	3.1	128 518 (1.8)
No	3 359 280 (98.2)	0.9	3 477 999 (98.1)	0.2	10 321 (87.7)	2.8	1 980 (96.7)	0.4	12 301 (89.0)	2.4	6 849 580 (98.2)
Sex while intoxicated and/or high on drugs
Yes	427 551 (12.5)	0.9	219 040 (6.2)	0.3	1 900 (16.1)	3.2	308 (15.0)	0.6	2 208 (16.0)	2.9	648 799 (9.3)
No	2 992 509 (87.5)	0.9	3 325 180 (93.8)	0.2	9 871 (83.9)	2.8	1 739 (85.0)	0.3	11 610 (84.0)	2.4	6 329 299 (90.7)
Sex with person of unknown HIV status
Yes	483 375 (14.1)	1.1	391 197 (11.0)	0.3	2 240 (19.0)	3.0	370 (18.1)	0.3	2 610 (18.9)	2.6	877 182 (12.6)
No	2 936 685 (85.9)	0.9	3 153 023 (89.0)	0.2	9 531 (81.0)	2.8	1 677 (81.9)	0.4	11 208 (81.1)	2.5	6 100 916 (87.4)
Sex with anonymous partner
Yes	187 941 (5.5)	1.3	64 024 (1.8)	0.4	1 195 (10.2)	3.1	143 (7.0)	1.4	1 338 (9.7)	2.9	253 303 (3.6)
No	3 232 119 (94.5)	0.9	3 480 196 (98.2)	0.2	10 576 (89.8)	2.8	1 904 (93.0)	0.3	12 480 (90.3)	2.4	6 724 795 (96.4)
Total	3 420 060 (100.0)	0.9	3 544 220 (100.0)	0.2	11 771 (100.0)	2.9	2 047 (100.0)	0.4	13 818 (100.0)	2.5	6 978 098 (100.0)

Note. FTM = female-to-male; IDU = injection drug use; MTF = male-to-female.

^aExcluding missing and invalid value for gender variable (37 980).

^bBecause of rounding and invalid or missing data, the values in each column may not sum to the column total.

Among transgender testing events, male-to-female transgender individuals reported higher frequencies than female-to-male transgender individuals for all risk behaviors, except IDU, during the past 12 months (male-to-female transgender individuals 5.7% vs female-to-male transgender individuals 6.3%).

Newly Identified, Confirmed HIV-Positive Testing Events

Among transgender testing events, 84.8% of the testing events and 97.0% of the newly identified confirmed HIV-positive testing events were conducted among male-to-female transgender individuals; therefore, we focused the statistical analysis on male-to-female transgender testing events that reported risk behavior (Table 3). Of the 336 newly identified confirmed HIV-positive male-to-female transgender testing events, 53.9% were among African Americans, followed by Hispanics/Latinos (29.8%), other racial/ethnic groups (9.5%), and Whites (6.8%). The highest percentage of newly identified confirmed HIV-positive testing events (53.0%) occurred among male-to-female transgender individuals aged 20 to 29 years, followed by male-to-female transgender individuals aged 30 to 39 years (25.0%). The highest percentage of newly identified confirmed HIV-positive testing events were from tests conducted in the South (35.4%), followed by the Northeast (30.7%) and the West (24.1%). The majority (61.3%) of newly identified confirmed HIV-positive testing events were conducted at non–health care facilities.

TABLE 3—

Characteristics Associated With a Newly Identified, Confirmed HIV-Positive Testing Event Among Male-to-Female Transgender Persons: United States 2009–2011

Characteristic	HIV Testing Event, No. (%)	Newly Identified, Confirmed HIV-Positive Testing Event, No. (%)	Bivariate Analysis, OR (95% CI)	Multivariable Analysis, AOR (95% CI)
Age at test, y
13–19	937 (8.0)	11 (3.3)	1.0 (Ref)	1.0 (Ref)
20–29	5 532 (47.0)	178 (53.0)	2.8 (1.5, 5.2)	2.8 (1.4, 5.6)
30–39	2 855 (24.3)	84 (25.0)	2.6 (1.4, 4.8)	2.5 (1.2, 5.0)
40–49	1 505 (12.8)	47 (14.0)	2.7 (1.4, 5.3)	2.8 (1.3, 5.9)
≥ 50	857 (7.3)	14 (4.2)	1.4 (0.6, 3.1)	1.6 (0.6, 3.9)
Race/ethnicity
African American	4 072 (34.6)	181 (53.9)	4.2 (2.7, 6.5)	4.6 (2.7, 7.9)
Hispanic/Latino	3 983 (33.8)	100 (29.8)	2.3 (1.5, 3.7)	2.6 (1.5, 4.5)
White	2 110 (17.9)	23 (6.8)	1.0 (Ref)	1.0 (Ref)
Other	1 600 (13.6)	32 (9.5)	1.9 (1.1, 3.2)	1.9 (1.0, 3.7)
Region
Northeast	3 676 (31.2)	103 (30.7)	1.0 (Ref)	NA
Midwest	1 157 (9.8)	27 (8.0)	0.8 (0.5, 1.3)
South	3 311 (28.1)	119 (35.4)	1.3 (1.0, 1.7)
West	3 431 (29.1)	81 (24.1)	0.8 (0.6, 1.1)
US dependent areas	196 (1.7)	6 (1.8)	1.1 (0.5, 2.5)
Test site type
Health care facilities	4 246 (36.1)	107 (31.8)	0.9 (0.7, 1.1)	NA
Non–health care facilities	7 005 (59.5)	206 (61.3)	1.0 (Ref)
Correctional facilities	278 (2.4)	15 (4.5)	1.9 (1.1, 3.2)
Other facilities	172 (1.5)	7 (2.1)	1.4 (0.6, 3.0)
Sex without using condom
Yes	7 241 (61.5)	225 (67.0)	1.9 (1.4, 2.6)	1.9 (1.3, 2.6)
No	2 706 (23.0)	45 (13.4)	1.0 (Ref)	1.0 (Ref)
Sex with IDU
Yes	628 (5.3)	23 (6.8)	1.3 (0.9, 2.0)	NA
No	9 133 (77.6)	255 (75.9)	1.0 (Ref)
Sex with person who is HIV positive
Yes	1 084 (9.2)	61 (18.2)	2.3 (1.7, 3.0)	1.5 (1.1, 2.0)
No	8 961 (76.1)	230 (68.5)	1.0 (Ref)	1.0 (Ref)
IDU in past 12 mo
Yes	670 (5.7)	31 (9.2)	1.7 (1.2, 2.5)	2.0 (1.3, 3.0)
No	9 899 (84.1)	273 (81.3)	1.0 (Ref)	1.0 (Ref)
Total	11 771 (100.0)	336 (100.0)

Note. AOR = adjusted odds ratio; CI = confidence interval; IDU = injection drug use; NA = not applicable; OR = odds ratio.

In the bivariate analysis, characteristics associated with newly identified, confirmed HIV-positive testing events among male-to-female transgender individuals included age (20–29, 30–39, and 40–49 years), race/ethnicity (African American and Hispanic/Latino), test site type (correctional facilities), and certain reported risk behaviors (sex without condom, sex with person who is HIV-positive, and IDU in past 12 months; Table 3). Among male-to-female transgender testing events, African American, Hispanic/Latino, and other racial/ethnic group had 4.2, 2.3, and 1.9 times the odds, respectively, of receiving a newly identified confirmed HIV-positive testing event compared with Whites. Testing events among male-to-female transgender individuals aged 20 to 29 had 2.8 times the odds, and those aged 40 to 49 and 30 to 39 years had 2.7 and 2.6 times the odds, respectively, of receiving a newly identified confirmed HIV-positive testing event compared with those aged 13 to 19 years. Testing events among male-to-female transgender individuals who reported sex without a condom, sex with a person who was HIV-positive, or IDU in the past 12 months had 1.9, 2.3, and 1.7 times the odds of receiving a newly identified confirmed HIV-positive testing event compared with those who did not report the behavior (Table 3).

In the multivariable model, age at test, race/ethnicity, sex without a condom, sex with a person who was HIV-positive, and IDU in the past 12 months all remained statistically significant based on a total of 8898 testing events (Table 3). Test site type and region were no longer significant when entered into the multivariable model. After adjustment for the remaining variables in the model, the characteristics most strongly associated with a newly identified, confirmed HIV-positive male-to-female transgender testing event included being 20 to 29 and 40 to 49 years old (adjusted odds ratio [AOR] = 2.8; 95% CI = 1.4, 5.6 and AOR = 2.8; 95% CI = 1.3, 5.9, respectively) compared with those aged 13 to 19 years, being African American (AOR = 4.6; 95% CI = 2.7, 7.9) or Hispanic/Latino (AOR = 2.6; 95% CI = 1.5, 4.5) compared with being White, reporting sex without a condom (AOR = 1.9; 95% CI = 1.3, 2.6) compared with reporting condom use, reporting sex with person who was HIV-positive (AOR = 1.5; 95% CI = 1.1, 2.0) compared with a person who was not, and reporting IDU in past 12 months (AOR = 2.0; 95% CI = 1.3, 3.0) compared with no IDU. The Hosmer and Lemeshow goodness-of-fit test was nonsignificant (χ² = 13.6; df = 9; P = .1378) and implied that the overall fit of the model was acceptable.

DISCUSSION

These testing data indicated that transgender testing events, although they represented a small percentage of the overall CDC-funded testing events, had higher levels of HIV positivity than male and female testing events. These data followed similar patterns documented in the literature on HIV testing and positivity among transgender populations and highlighted that a high proportion of transgender individuals are HIV-infected.³

Among transgender testing events, male-to-female transgender individuals represented the majority of testing events, and, consistent with other reports, nearly all of the HIV-positive testing events (97%).^3,9,10 Female-to-male transgender individuals reported a high frequency of risk behaviors; however, consistent with other findings, only 12 HIV-positive tests (0.5%) were identified in these data.^10,15–17 Younger (aged 20–39 years), African American, and Hispanic/Latino transgender individuals had the highest percentages of HIV positivity. It was possible that lower percentages of positivity among transgender adolescents aged 13 to 19 years reflected some of the complexities of accurately surveying transgender adolescents.⁴⁶ In addition, in the multivariable model, being African American or Hispanic/Latino in comparison with White was a strong predictor of a newly identified confirmed HIV-positive testing event among male-to-female transgender individuals. This suggested that African American and Hispanic/Latino male-to-female transgender individuals were disproportionately affected by HIV; prevention programming is needed for these populations.

More testing events were conducted in non–health care settings for transgender individuals than for male and female individuals, possibly highlighting barriers to accessing HIV testing in traditional health care settings, such as lack of health insurance,¹⁰ perceived provider insensitivity,^11,47 or the perception that the site is not “trans-friendly.”⁴⁸ Based on these data, the transgender population accessed HIV testing mostly through local community-based organizations, or other non–health care testing venues, highlighting the importance of incorporating HIV testing into transgender-focused programs and in nontraditional health care settings. Male-to-female transgender individuals in correctional facilities appeared to have a higher positivity rate than any other gender category (5.1%), which is a concern (Table 1). The elevated prevalence of HIV positivity in correctional settings might be explained by high levels of HIV risk behaviors (e.g., sexual violence, risky sexual behaviors, no condom use) within these facilities or by the characteristics of this population before incarceration.^49,50 Transgender individuals might be particularly vulnerable in the correctional facility context, where they might not be able to freely identify as transgender, and might experience risky situations or behaviors, which could potentially result in increased risk of HIV. These data indicated the need for better HIV prevention interventions that target male-to-female transgender individuals in correctional facilities.

Transgender individuals (regardless of female-to-male or male-to-female) had a higher percentage of receiving their test results than male and female individuals, which might indicate that testing facilities where transgender individuals receive results are efficient in returning results, or that the transgender population is motivated to receive their results. Programs that target the transgender population and incorporate HIV testing into their services might find a high level of receptivity to HIV testing among their clients. In addition, targeting HIV prevention and testing activities to the transgender population in non–health care settings might work to identify undiagnosed infections that otherwise would not have been found in traditional health care settings.

Consistent with the literature, testing events among male-to-female transgender individuals indicated a higher frequency of most risk behaviors in comparison with other genders (Table 2).^3,10 Many transgender individuals face stigma and discrimination by employers; the resulting financial vulnerability may push some to sex work to earn their living or support use of street drugs, hormones, or silicone.^10,11,24 HIV prevention programs that target male-to-female transgender individuals might consider promoting training for employers around job site discrimination.^51–53 Furthermore, HIV prevention programs could place a special focus on risk-reduction strategies as a way to decrease risky behaviors among high-risk transgender individuals. IDU was more frequently reported by transgender individuals compared with male and female individuals. This might be partially caused by injecting of hormones or street drugs (we did not collect detailed information on type of injecting drug), indicating the importance of incorporating education on safe injection methods and substance abuse prevention into prevention programming.^2,3,10,11,24

Limitations

Our findings in this article were subject to some limitations. First, the estimated proportion of persons tested for HIV in publicly funded sites was only between 15% and 20% of all persons tested in United States in the past 12 months.⁵⁴ Thus, HIV testing in CDC-funded sites was not representative of all testing in the United States, and interpretation from a national perspective was limited.

Second, our unit of analysis was testing events rather than tested persons. In the absence of a standard data collection system across agencies and project areas, it was difficult to ascertain the degree to which a testing event represented a person. However, we created a plausible unique person identifier in the data set by combining 4 existing variables that were unduplicated for 93.7% of the HIV testing events, implying the high likelihood that an HIV testing event represented a unique person. The unique person identifier might not be able to ascertain if individuals repeated tests across sites or jurisdictions, but we assumed that these numbers were relatively small. Although we had no estimate of how closely the number of testing events matched the number of tested persons, it was reasonable to assume that testing events were an adequate proxy for the number of tested persons.

Third, with test-level data, it was not possible to link the results of repeat tests for the same person. However, the definition of a newly identified confirmed HIV-positive testing event used in our analysis minimized this limitation for persons who were newly identified, because records for which there was a current confirmed HIV-positive testing event and a history of a HIV-positive testing event were excluded. We found that the proportion of a self-reported previous test was higher among male-to-female transgender individuals (77.7%) and female-to-male transgender individuals (71.0%) than it was among male (53.5%) and female (51.1%) individuals, which implied that transgender individuals were more likely to have a history of testing and that they might be overly represented in these data. This was a potential source of bias in estimates about testing rates among various gender groups, which was not a primary focus of our study. Instead, we focused on HIV positivity rates and factors associated with positivity among transgender individuals. Because our definition of newly identified confirmed HIV-positive testing events accounted for previous HIV testing and seroconversion during a recent test event, the bias associated with repeat testing was likely to be minimal.

Fourth, because of exclusion criteria, the analytical sample was less likely to contain HIV testing events from Whites, males, Northeast and Midwestern regions, and those conducted in health care settings. However, the bias introduced in our analysis was likely to be minimal because of the small proportion of HIV testing events that were excluded from analysis (0.8%).

Fifth, these data were collected through HIV prevention programs, and the comparability of data across health departments might be limited because of differences in data collection and quality assurance activities that occur at state or local levels. These data underwent CDC data quality and control procedures.

Lastly, we collected transgender data using a 1-step, self-reported answer to the question “please indicate your current gender identity” rather than a 2-step question model (i.e., what was your assigned gender at birth and what is your current gender identity). The 1-step model might underestimate the transgender population,⁵⁵ and thus, it was possible that there were additional transgender testing events included in the male and female gender categories. In 2012, CDC instituted requirements for reporting HIV testing data using the 2-step model to better capture gender identity and monitor participation in prevention programs among gender minority groups.⁵⁶ This 2-step approach has since been widely recommended by experts, with special considerations with regard to age (particularly adolescents), race/ethnicity, and other factors.⁴⁶ Future analyses are needed to examine the degree of misclassification of transgender groups and the effect misclassification may have on estimates of HIV positivity.

Conclusions

Analyses of testing events from CDC-funded sites revealed high levels of HIV positivity among transgender individuals, particularly male-to-female transgender individuals. Among male-to-female transgender individuals, African American and Hispanic/Latino male-to-female transgender individuals were particularly affected. Our findings underscored the need for expanding targeted HIV prevention services that are responsive to the needs and sociodemographic characteristics of transgender individuals. Prevention efforts could focus on non–health care settings as a way of reaching transgender individuals with prevention messages. In addition, our findings highlighted the need to improve HIV data collection and surveillance systems by including a more accurate method for identifying transgender individuals to better understand the dynamics of HIV within this at-risk population.