Viral Suppression and HIV Transmission Behaviors among Hospitalized Patients Living with HIV

Abstract

From July 2012 to January 2014, the Project HOPE study interviewed 1,227 people with HIV infection from 11 hospitals in the US to determine eligibility for participation in a randomized trial. Using these screening interviews, we conducted a cross-sectional study with multivariable analysis to examine groups that are at highest risk for having a detectable viral load and engaging in HIV transmission behaviors. Viral suppression was 42.8%. Persons with a detectable viral load were more likely to have sex partners who were HIV-negative or of unknown status (OR=1.72, 95% CI=1.22-2.38), report not cleaning needles after injecting drugs (OR=3.13, 95% CI=1.33-7.14), and to engage in sex acts while high on drugs or alcohol (OR=1.85, 95% CI=1.28-2.7) compared to their counterparts. Many hospitalized people with HIV infection are unsuppressed and more likely to engage in HIV transmission behaviors than those with viral suppression. Developing behavioral interventions targeting HIV transmission behaviors toward patients with unsuppressed HIV viral load in the hospital settings has the potential to prevent HIV transmission.

1. Introduction

There are more than 1.14 million people with HIV infection in the United States (US) at the end of 2016, with approximately 39,000 new diagnoses in 2017 ^{1, 2}. In 2016, gay and bisexual men accounted for 67% of all HIV diagnoses and 82% of diagnoses among males aged 13 and older ³. Women made up 19% of the new HIV diagnoses in US in 2016, of which about 87% were from heterosexual contact while 12% were from injection drug use (IDU) ⁴. Sexual transmission is the most common cause of HIV transmission in the US ².

To maintain their individual health and limit further transmission, people with HIV infection should seek HIV care and take antiretroviral therapy to treat HIV as soon as possible to achieve an undetectable viral load. People with HIV infection who reduce the amount of virus also help prevent transmission to others through sex or syringe sharing, referred to as “treatment as prevention” ⁵. If viral load is very low or undetectable, the chance of transmitting HIV is greatly reduced ⁶. In fact, people with HIV infection whose virus is completely, durably suppressed by treatment will effectively not sexually transmit the virus, known as the Undetectable=Untransmittable (U=U) HIV prevention strategy ^{7, 8}.

The most recent statistics show that only 52% of gay and bisexual men with HIV infection and 48% of women with HIV infection were virally suppressed in 2014–2015 ^{3, 4}. Approximately 37% of people with HIV infection in the US were not receiving care and taking antiretroviral therapy (ART) and 51% were not retained or engaged in care in 2015 ⁹. In addition, these late and intermittent treatment seekers are more likely to be illicit drug users who frequently fail to follow up in HIV outpatient clinics ^10-12. Because patients not treated with ART are more likely to have high viral loads and more infectious fluids, there is increased risk of transmission to their seronegative sex and drug using partners ¹³. Therefore, an understanding of whether and how HIV transmission risk behaviors vary across people with HIV infection by viral suppression status (suppressed vs. unsuppressed) could provide important information for health providers who may need to intervene with particular subpopulations, not only for care but for prevention purposes.

Little research has examined how behavioral risk profiles differ by HIV viral suppression status among inpatients in medical care settings, including for hospitalized people with HIV infection. Viral suppression improves health outcomes for people with HIV infection and reduces the risk of HIV transmission at the population level ^{14, 15}. This study uses screening data from a randomized controlled trial to describe the association between viral suppression and transmission behavior characteristics, particularly injection drug use and sexual risk behaviors, among hospitalized people with HIV infection in the US. We hypothesize that 1) there will be a difference between socio-demographics and gender and sexual orientation among the hospitalized people with HIV infection; 2) there will be differences by gender and sexual orientation in reported injection and sexual risk behaviors; 3) there will be differences in the associations between socio-demographic factors and viral suppression status (suppressed vs. unsuppressed) by gender and sexual orientation.

2. Methods

2.1. Participants

Between July 2012 and January 2014, 2,291 hospitalized people with HIV infection were interviewed at bedside as part of the screening assessment for the CTN0049 study, Project HOPE (Hospital Visit as Opportunity for Prevention and Engagement for HIV-infected Drug Users). Recruitment occurred in 11 US hospitals (Atlanta, Georgia; Baltimore, Maryland; Boston, Massachusetts; Birmingham, Alabama; Chicago, Illinois; Dallas, Texas; Los Angeles, California; Miami, Florida; New York, New York; Philadelphia and Pittsburgh, Pennsylvania). All people with HIV infection at the hospitals were approached provided staff clinicians deemed them cognitively and medically able to participate in the screening procedures and verified their interest in participating in research. Individuals were excluded if they were prisoners (any inpatient who was currently in jail, prison or any inpatient overnight facility as required by a court of law, or any inpatient who was currently awaiting trial, on probation or under house arrest, and was escorted to the hospital from a jail, prison or any inpatient overnight facility as required by law, or would be escorted to such a facility upon discharge as required by law), under 18 years of age, unable to provide informed consent and basic contact information for follow-up or unable to communicate in English. To be included, patients needed to have detectable HIV viremia (i.e., >200 copies/mL) or unknown viral load and a CD4 count </=500/uL in the past 12 months. An “unknown” viral load was accepted if accompanied by the Site principal investigators’ discretion that the patient a) is likely to currently have a viral load >200 copies/mL, b) is not currently successfully/correctly taking ART and c) needs to be on ART. While all sites began by screening all people with HIV infection in the hospital, a number of sites initiated pre-screening medical records at a later stage of the Project HOPE study ¹⁶. Participants included in the present analysis are those recruited prior to the implementation of the updated pre-screening process (N=1,227) as to represent all patients admitted to the hospital during this time who were medically cleared and who agreed to screen for the study. The 413 patients who met the “unknown” viral load criteria were excluded from the viral suppression analysis. The study was approved by local Institutional Review Boards at each participating institution.

2.2. Data collection and Measures

Computer-assisted personal interviews (CAPIs) were used to obtain participants’ self-reported behaviors in the 12 months before the interview, unless otherwise noted. Men were classified as men who have sex with men (MSM) if they reported sex with a male partner in the past 12 months. Men were classified as men who have sex with women (MSW) if they reported sex with women exclusively in the past 12 months. All women, regardless of sexual behavior, were categorized as one group due to the small numbers of women who had sex with women (reported sex with women exclusively in the past 12 months, N=7, 1.8%) and the lower risk of HIV transmission between women. Other socio-demographic characteristics collected included: age, race/ethnicity (non-Hispanic White, non-Hispanic Black, Hispanic, other), unstable housing (homeless, in a shelter, room occupancy hotel, HIV/AIDS housing/group home, drug treatment facility, other residential facility or institution etc. within the past six months), food insecurity (enough to eat, sometimes not enough, and often not enough within the past four weeks), education (less than high school, high school, and more than high school), annual income ($0–20,000, $20,001–40,000, and greater than $40,001), incarceration history (ever in jail, and in jail in the last 6 months), health insurance status (have health insurance or not), and IDU (ever IDU, and IDU in the last 12 months). Viral suppression was defined as having a viral load </= 200 copies/ml in the past 12 months. The cutoff value of </= 200 HIV RNA copies/ml was based on the US Department of Health and Human Services recommended definition of virologic failure at the time of the study (i.e., failure of antiretroviral therapy to suppress a viral load to </= 200 copies/mL).

IDU and sexual risk behavior questionnaires were adapted from the Global Appraisal of Individual Needs (GAIN) ^17-19. Questions included the use of a needle for injection with drugs or medication during the past 12 months, including injection by someone besides a doctor or nurse or self-injected prescribed medication, and needle cleaning/sharing behaviors. For example, during the past 12 months, did you… a) Use a needle to shoot up drugs? b) Reuse a needle that you had used before? c) Reuse a needle without cleaning it with bleach or boiling water first? d) Use a needle that you knew or suspected someone else had used before? e) Use someone else’s rinse water, cooker, or cotton after they did? f) Ever skip cleaning your needle with bleach or boiling water after you were done? g) Let someone else use a needle after you used it? h) Let someone else use the rinse water, cooker, or cotton after you did? i) Allow someone else to inject you with drugs?

Sexual risk questions assessed whether an individual had vaginal and/or anal sex, and condom use with anyone during the past 12 months. For example, during the past 12 months, did you… a) Have sex while you were high on alcohol or on other drugs? b) Have sex with someone who was an injection drug user? c) Have sex involving anal intercourse (penis to butt)? d) Have sex against your will (you were forced or coerced)? e) Trade sex to get drugs, gifts or money? f) Use drugs, gifts or money to purchase or get sex? g) Have sex with someone who you thought was HIV negative or you did not know their HIV status? h) Have two or more different sex partners (not necessarily at the same time)? i) Have sex without using any kind of condom to protect you and your partner from diseases or pregnancy? j) Have a lot of pain during sex or after having had sex? k) Use alcohol or other drugs to make sex last longer or hurt less?

2.3. Statistical Analysis

SAS 9.3 (SAS Institute, Cary, NC, USA) was used for all data analyses. One participant did not provide their gender/sexual orientation status and was therefore excluded. Descriptive statistics were used to characterize participants’ socio-demographic information, and the prevalence of IDU and sexual risk behaviors was assessed separately by the gender/sexual orientation status of MSW, women (including both women having sex with men and women), and MSM. Chi-squared tests of independence and ANOVAs were used to evaluate associations and statistical significance for categorical variables and continuous variables with gender/sexual orientation status, respectively. Fisher’s exact test was used if the cells had an expected frequency of five or less. We used multiple logistic regression to calculate odds ratios and 95% confidence intervals (CIs) for the association between viral suppression as an independent variable and IDU and sexual risk behaviors as dependent variables, controlling for age and race/ethnicity based on previous literature ^20-29. P-values < .05 for two-sided tests were considered statistically significant. To protect against the false discovery rate due to multiple comparisons, probability values were also checked using the Benjamini–Hochberg procedure ³⁰.

3. Results

Among the 1,226 people with HIV infection in the analysis, 629 (51.3%) were MSW, 389 (31.7%) were women, and 208 (17.0%) were MSM (Table 1). There were statistically significant differences by gender/sexual orientation (MSW, women and MSM) across all socio-demographic characteristics except for incarceration in the last 6 months. Specifically, MSM were more likely than women and MSW to be White (MSM=25.1%, women=9.8% and MSW=16.8%; p<.0001), to have more than a high school education (MSM=54.8%, women=26.8% and MSW=32.6%, p<.0001), and to have higher incomes (MSM=39.1%, women=19.6% and MSW=30.2% in $20,001-$40,000; p<.0001). In addition, MSM were less likely to have health insurance (MSM=68.3%, women=85.9% and MSW=80.2%; p<.001). In contrast, MSW were more likely than women and MSM to report unstable housing (MSW=32.7%, women=23.2%, and MSM=22.1%, p=.001), food insecurity (often not enough food: MSW=11.7%, women=6.5%, and MSM=7.7%, p=.048), and incarceration (ever in jail: MSW=68.3%, women=59.4%, and MSM=55.8%, p=.001). There were no significant differences between gender/sexual orientation and ever IDU (MSW=36.3%, women=35.7%, and MSM=28.4%; p=.102), or viral suppression (MSW=42.2%, women=44.1%, and MSM=41.9%; p=.861). Overall, 42.8% of individuals were virally suppressed.

Table 1.

Sociodemographic and Clinical Characteristics of Adults with HIV Diagnosis for the CTN0049 Study, 11 Hospitals (Atlanta, Georgia; Baltimore, Maryland; Boston, Massachusetts; Birmingham, Alabama; Chicago, Illinois; Dallas, Texas; Los Angeles, California; Miami, Florida; New York, New York; Philadelphia and Pittsburgh, Pennsylvania), United States, 2012-2014

Characteristics		MSW		Women		MSM		Total
		N=629	%	N=389	%	N=208	%	N=1226	%	Chisq	df	p-value
Age (mean, sd)		49.3 (9.9)		45.7 (10.3)		40.7 (10.3)		46.7 (10.6)		120.1	1	<0.001
Race	Hispanic	72/620	11.6	33/386	8.5	37/207	17.9	142/1213	11.7	44.1	6	<0.001
	Black	413/620	66.6	288/386	74.6	104/207	50.2	805/1213	66.4
	White	104/620	16.8	38/386	9.8	52/207	25.1	194/1213	16
	Other	31/620	5.0	27/386	7.0	14/207	6.8	72/1213	5.9
Unstable housing		199/609	32.7	88/379	23.2	46/208	22.1	333/1196	27.8	14.5	2	0.001
Food insecurity	Enough to eat	445/623	71.4	284/386	73.6	150/208	72.1	879/1217	72.2	9.6	4	0.048
	Sometimes not enough	105/623	16.9	77/386	19.9	42/208	20.2	224/1217	18.4
	Often not enough	73/623	11.7	25/386	6.5	16/208	7.7	114/1217	9.4
Education	Less than high school	198/624	31.8	141/388	36.3	31/208	14.9	370/1220	30.3	55.3	4	<0.001
	High school	222/624	35.6	143/388	36.9	63/208	30.3	428/1220	35.1
	More than high school	203/624	32.6	104/388	26.8	114/208	54.8	421/1220	34.6
Income	$0-$20,000	326/497	65.6	229/296	77.4	82/161	50.9	637/954	66.8	37.3	4	<0.001
	$20,001-$40,000	150/497	30.2	58/296	19.6	63/161	39.1	271/954	28.4
	$40,001 or more	21/497	4.2	9/296	3	16/161	9.9	46/954	4.8
Incarceration	Ever in jail	424/621	68.3	230/387	59.4	116/208	55.8	770/1216	63.3	14.2	2	0.001
	Last 6 month in jail	62/424	14.6	26/230	11.3	19/116	16.4	107/770	13.9	2.1	2	0.354
Health insurance		498/621	80.2	328/382	85.9	142/208	68.3	968/1211	79.9	26.0	2	<0.001
IDU	Ever IDU	226/622	36.3	137/384	35.7	59/208	28.4	422/1214	34.8	4.6	2	0.102
Viral suppression* (</=200 copies/ml)		173/410	42.2	113/256	44.1	62/148	41.9	348/814	42.8	0.3	2	0.861

^*413 met the “unknown” viral load criteria and therefore were excluded from the viral suppression analysis.

Among the sample as a whole, 11.4% used needles to inject drugs in the past 12 months (Table 2) and 6.6% reused their own needles; the prevalence of other needle-related risk behaviors is shown in Table 2. Only two IDU-related behaviors differed by gender/sexual orientation. MSM were less likely than women or MSW to report that “someone uses a needle after you” (MSM=1%, women=4.7%, and MSW=2.6%; p=.028) and that “someone uses a rinse/cooker/cotton after you (MSM=0, women=4.7%, and MSW=2.1%; p=.001).

Table 2.

Injection Drug Use and Sexual Risk Behaviors in the Last 12 Months of MSW, Women and MSM, for the CTN0049 Study, 11 Hospitals (Atlanta, Georgia; Baltimore, Maryland; Boston, Massachusetts; Birmingham, Alabama; Chicago, Illinois; Dallas, Texas; Los Angeles, California; Miami, Florida; New York, New York; Philadelphia and Pittsburgh, Pennsylvania), United States, 2012-2014

Characteristics	MSW		Women		MSM		Total
	N=629	%	N=389	%	N=208	%	N=1226	%	Chisq	df	p-value
IDU in the last 12 months
Use needle to shoot up	76/622	12.2	39/384	10.2	23/208	11.1	138/1214	11.4	1.0	2	0.599
Reuse needle you used	42/622	6.8	23/384	6	15/208	7.2	80/1214	6.6	0.4	2	0.826
Reuse needle without cleaning	21/622	3.4	13/384	3.4	10/208	4.8	44/1214	3.6	1.0	2	0.605
Reuse other's needle	17/622	2.7	13/384	3.4	4/208	1.9	34/1214	2.8	1.1	2	0.615
Use other's rinse/cooker/cotton	23/622	3.7	17/384	4.4	2/208	1	42/1214	3.5	5.1	2	0.057
Not cleaning needle after injecting drugs	32/622	5.1	18/384	4.7	11/208	5.3	61/1214	5	0.1	2	0.932
Someone uses needle after you	16/622	2.6	18/384	4.7	2/208	1	36/1214	3	7.2	2	0.028
Someone uses rinse/cooker/cotton after you	13/622	2.1	18/384	4.7	0/208	0	31/1214	2.6	13.0	2	0.001
Allow someone inject you	25/622	4	16/384	4.2	10/208	4.8	51/1214	4.2	0.2	2	0.886
Sexual risk behavior in the last 12 months
Have sex high on alcohol/drugs	90/613	14.7	64/382	16.8	98/208	47.1	252/1203	20.9	104.6	2	<0.001
Have sex with IDU user	43/611	7	23/380	6.1	28/207	13.5	94/1198	7.8	11.5	2	0.003
Have anal sex intercourse	11/614	1.8	18/381	4.7	157/208	75.5	186/1203	15.5	694.6	2	<0.001
Have sex against will	8/614	1.3	17/382	4.5	9/208	4.3	34/1204	2.8	10.6	2	0.005
Trade sex to get drugs/gifts/money	7/614	1.1	36/382	9.4	24/208	11.5	67/1204	5.6	47.8	2	<0.001
Use drugs/gifts/money to purchase sex	26/614	4.2	15/382	3.9	21/208	10.1	62/1204	5.1	12.6	2	0.002
Have sex with HIV negative/unknown status partner	97/614	15.8	91/381	23.9	104/208	50	292/1203	24.3	98.9	2	<0.001
Multiple partners	59/614	9.6	50/382	13.1	95/208	45.7	204/1204	16.9	149.5	2	<0.001
Condomless vaginal or anal sex	96/614	15.6	86/382	22.5	114/208	54.8	296/1204	24.6	129.9	2	<0.001
Have pain during/after sex	4/614	0.7	32/382	8.4	14/208	6.7	50/1204	4.2	39.5	2	<0.001
Use alcohol/drugs to make sex longer/hurt less	42/614	6.8	20/382	5.2	44/208	21.2	106/1204	8.8	48.5	2	<0.001

There were significant differences among gender/sexual orientation for prevalence of engaging in sex while high on alcohol or drugs (Table 2). Particularly, MSM were significantly more likely to engage in sex while high on alcohol or drugs, condomless vaginal or anal sex, sex with HIV negative/unknown status partners, and multiple partners, compared to MSW or women.

Virally unsuppressed individuals were less likely than those who were suppressed to report having ever injected drugs when controlling for age and race/ethnicity (30.5% versus 39.7%, OR=0.67, 95% CI=0.50–0.89, p=.007) (Table 3). IDU or needle sharing in the last 12 months were not significantly higher among virally unsuppressed individuals compared to their counterparts (p=0.067). However, virally unsuppressed participants were more likely to not clean needles after injecting drugs (OR=3.13, 95% CI=1.33–7.14, p=.009), to engage in sex acts while high on drugs or alcohol (OR=1.85, 95% CI=1.28–2.7, p=.001), and to engage in sex with partners who were HIV-negative or of unknown serostatus (OR=1.72, 95% CI=1.22–2.38, p=.002).

Table 3.

Logistic Regression for Viral Suppression (ref=viral suppressed), for the CTN0049 Study, 11 Hospitals (Atlanta, Georgia; Baltimore, Maryland; Boston, Massachusetts; Birmingham, Alabama; Chicago, Illinois; Dallas, Texas; Los Angeles, California; Miami, Florida; New York, New York; Philadelphia and Pittsburgh, Pennsylvania), United States, 2012-2014

Characteristics	Unsuppressed %	Suppressed %	OR (95% CI)	p-value
Ever IDU	30.5	39.7	0.67 (0.50, 0.89)	0.007
IDU in the last 12 months
Use needle to shoot up	10.8	7.0	1.61 (0.97, 2.70)	0.067
Reuse needle you used	6.5	4.4	1.52 (0.80, 2.86)	0.201
Reuse needle without cleaning	3.9	2.3	1.69 (0.73, 4.00)	0.222
Reuse other's needle	2.2	1.7	1.23 (0.45, 3.45)	0.68
Use other's rinse/cooker/cotton	3.0	2.9	1.04 (0.45, 2.38)	0.929
Not cleaning needle after injecting drugs	6.0	2.0	3.13 (1.33, 7.14)	0.009
Someone uses needle after you	2.8	2.0	1.39 (0.55, 3.45)	0.491
Someone uses rinse/cooker/cotton after you	2.4	1.5	1.64 (0.56, 4.76)	0.36
Allow someone inject you	3.7	2.3	1.59 (0.68, 3.70)	0.282
Sexual Risk Behavior in the last 12 months
Have sex high on alcohol/drugs	23.9	14.4	1.85 (1.28, 2.70)	0.001
Have sex with IDU user	8.3	5.9	1.43 (0.82, 2.50)	0.207
Have anal sex intercourse	16.5	14.7	1.15 (0.78, 1.69)	0.483
Have sex against will	3.3	1.8	1.89 (0.72, 5.00)	0.197
Trade sex to get drugs/gifts/money	5.6	2.9	1.96 (0.94, 4.17)	0.072
Use drugs/gifts/money to purchase sex	5.2	3.2	1.64 (0.79, 3.45)	0.179
Have sex with HIV negative/unknown status partner	28.0	18.5	1.72 (1.22, 2.38)	0.002
Multiple partners	18.9	14.7	1.35 (0.93, 1.96)	0.118
Condomless vaginal or anal sex	25.6	23.5	1.12 (0.81, 1.56)	0.488
Have pain during/after sex	4.3	3.2	1.35 (0.64, 2.86)	0.421
Use alcohol/drugs to make sex longer/hurt less	9.1	7.0	1.32 (0.79, 2.22)	0.292

^*Controlled for age and race/ethnicity

4. Discussion

Over half (57%) of study participants recruited from the 11 hospitals were not virally suppressed, similar to the 59.8% estimate from Centers for Disease Control and Prevention (CDC) among the people with HIV infection nationwide in 2015 ³¹. Although we did not find a significant association between gender/sexual orientation and viral suppression, we identified relationships and differences between HIV transmission risk behaviors by gender/sexual orientation, and by whether an individual’s viral load was undetectable/detectable. Approximately one-quarter of the virally unsuppressed individuals engaged in two specific sexual risk behaviors: having sex while high on alcohol or drugs, and having sex with partners who were HIV-negative or of unknown serostatus. Each risk behavior was significantly higher compared to their virally suppressed counterparts. These findings are consistent with Mattson et al. ³², who found that virally unsuppressed people with HIV infection receiving medical care in the US were more likely than their virally suppressed counterparts to engage in vaginal or anal sex, unprotected vaginal or anal sex, or unprotected vaginal or anal sex with a partner of negative or unknown HIV status ³². However, our findings differ from two meta-analyses that found no association between viral suppression and sexual risk behaviors among people with HIV infection in the US ^{33, 34}. Our findings that virally unsuppressed individuals were significantly more likely to have sex while high, and with a partner of negative or unknown serostatus are of significant concern. These results demonstrate the need for implementing existing efficacious preventive interventions specifically targeted toward virally unsuppressed individuals, not only to provide medications to improve their health, but also to reduce sex while high and increase condom use with unknown serostatus partners. For example, Crepaz et al. ³⁵ summarized 14 evidence-based interventions in a systematic review of interventions for reducing HIV risk behaviors among people with HIV infection in the US from 1988–2012. All of the included interventions had low risk of bias and showed significant positive intervention effects on reducing HIV transmission risk behaviors.

More than half of the MSM with HIV infection in our sample reported condomless vaginal or anal intercourse, and half reported anal intercourse with a partner of negative or unknown HIV status. Although many sexual risk reduction interventions exist for MSM, high HIV prevalence and incidence rates still persist among this population; further implementation of MSM-specific interventions to address the factors associated with sexual risk behaviors is needed ^36-38. In addition, MSM are the consistent focus of HIV interventions, while fewer resources are focused on HIV prevention for individuals of other gender/sexual orientations. Despite effective prevention strategies such as treatment as prevention and pre-exposure prophylaxis (PrEP), the effectiveness of these biological approaches to reducing HIV transmission is challenged by systems- and structural-level factors that contribute to low health-seeking behaviors in some MSM ³⁹, as well as in other vulnerable populations, such as transgender women, people of color with HIV infection, and sex workers. ^40-43.

There were no significant differences in IDU-related risk behaviors between MSW and women. Among those who did report injecting drugs, MSM were less likely than MSW or women to report sharing their used needles, or using rinse, cookers or cotton, with others. These results were contradictory to a previous study on gender differences of IDU risk behaviors that found no gender difference for needle sharing behaviors ⁴⁴. However, another study showed that female IDUs were more likely than men to share injecting equipment with their sexual partners ⁴⁵. In addition, the majority of women who shared injecting equipment with their partners injected with their partners’ used needles rather than injecting first ⁴⁶. Another study reported increased risks for women who report a combination of sexual and injecting risk practices compared to men ⁴⁷.

Finally, the study found significant differences in socio-demographics among MSW, women and MSM with HIV infection. First, the MSM in this sample were younger than the MSW or women, which may, in part, explain the level of sexual activity, and high transmission category through male-to-male sex contact among young MSM. In fact, previous research has demonstrated that young MSM are more likely to acquire HIV through sexual intercourse compared to MSW or women ^{48, 49}. In 2016, 82% of HIV infections diagnosed among male adults and adolescents were attributed to male-to-male sexual contact ⁴⁹. In addition, the diagnosed infections attributed to male-to-male sexual contact accounted for 92% of diagnosed HIV infections among male adolescents and 91% of diagnosed HIV infections among male young adults ⁴⁸. Second, we found that significantly more MSW were unstably housed, had ever been incarcerated, and/or injected drugs compared to MSM and women. On the other hand, MSM reported higher education and income, but were also less likely to have been incarcerated, compared to MSW and women. It is also striking that, among our sample of hospitalized people with HIV infection, MSM were most likely to lack health insurance compared to women or MSW. These disparities support the continued need to look beyond individual-level factors to explain differences in HIV infection ^50-53. Interventions focusing on individual-level behaviors, particularly for those who are out of care, need to be paired with community-level and structural interventions that address social and economic disparities associated with HIV infection. For example, support for adherence and retention in care, individualized risk assessment and counseling, assistance with partner notification, and periodic screening for common sexually transmitted infections are recommended for people with HIV infection as part of care ⁵⁴. Furthermore, additional patient support services such as patient health navigation, community and peer outreach, provision of culturally appropriate print media, verbal messages from clinic staff promoting health care utilization and retention, and youth-focused case management and support interventions should be implemented at both individual and structural levels to promote movement of people with HIV infection through the continuum of HIV care ⁵⁴.

While interventions could certainly be delivered in hospitalized settings, there is not much literature addressing how interventions would be tailored to this specific setting. Rather, the existing literature describes interventions being implemented in broader HIV treatment/care settings (i.e., clinics or outpatient sites). For example, two recent studies implemented counseling and behavioral interventions in outpatient centers and HIV care settings, but these only address high-risk sexual behaviors, not IDU behavior ^{55, 56}. Conclusions regarding efficacy in reduction of the high-risk sexual behaviors were limited ^{55, 56}, but that should not prevent such interventions or derivatives of such interventions from being tested in the hospital setting. Another more recent study used an Information-Motivation-Behavioral Skills model to examine the influence of neurocognitive impairment on reducing risk behaviors in drug users ⁵⁷. Although the drug users who participated were not HIV-infected, it would be worth extending the aforementioned model to address the influence of neurocognitive impairment on HIV risk reduction outcomes and to be applied to future interventions regarding high-risk drug users, or people with HIV infection in the hospital setting if adjusted accordingly.

Our findings should be interpreted in light of several limitations. First, despite the multisite design and large sample size, participants were not randomly selected, and therefore findings are not generalizable to the broader US population. Second, no causality could be inferred due to the cross-sectional study design. Third, individual characteristics (except for viral load which is based on medical record review or provider discretion/medical opinion) and behaviors were determined on the basis of self-report and may be subject to social desirability and other response biases. Finally, this is a secondary data analysis and should be viewed as exploratory.

In conclusion, our findings show that hospitalized patients living with HIV who are not virally suppressed reported more behaviors with a higher risk of HIV transmission than did those who are suppressed. This highlights the need for effective prevention interventions among people with HIV infection who are not virally suppressed, particularly MSM who reported a higher prevalence of sexual risk behaviors than MSW or women. Disparities in access to resources and healthcare highlight the importance of implementing broad-based interventions aimed at multiple structural-level factors as well as implementing and delivering individual-level HIV transmission behavioral risk reduction interventions for people with HIV infection who are not virally suppressed.