Lessons from +LOVE,Improving HIV Care for Black Men Who Have Sex with Men in New Orleans Using an Integrated Behavioral Health Crisis Support Model of Care

Abstract

Black men who have sex with men (BMSM) experience higher rates of HIV infection than other groups. They also face disparities in linkage and retention to HIV care, as well as in viral suppression. To address the needs of the BMSM population living with HIV, we developed a novel intervention program (+LOVE), which integrated case management with behavioral health/crisis support. The intervention consisted of (1) a dedicated therapist; (2) an SMS (text message)-based engagement program, CareSignal that provided medication reminders and administered routine ecological momentary assessments (EMAs); (3) a triage crisis coordinator to respond to alerts generated by the EMA responses; and (4) a case manager. This study assessed the additional impact of the first three components of the intervention (a dedicated therapist, text messaging application, along with a triage crisis coordinator), compared with the fourth component alone, the current standard of care (a case manager) on access to HIV care, antiretroviral therapy (ART) adherence, and viral suppression. Analyzing electronic medical record data, survey data, and EMA response data extracted from CareSignal, we found that those who engaged in the behavioral health therapy had higher odds of remaining in HIV care than those who only engaged with the case manager. We also found that increased engagement with CareSignal led to an increased probability of achieving positive HIV-related health outcomes. Our results suggest that an integrated behavioral health/crisis support model of intervention integrated with case managementincreased positive outcomes over case management alone.

Introduction

HIV infection rates are high in Black men who have sex with men (BMSM) in New Orleans. The New Orleans metropolitan statistical area (MSA) ranked sixth in the nation for HIV case rates among large metropolitan areas in 2019.¹ Among these new diagnoses, 66% were Black, although they comprise only 32% of the general population in the MSA.² The primary exposure category for all people with HIV (PWH) in New Orleans in 2019 was BMSM, at 71% of all PLWH. In 2019, of 250 new HIV cases in New Orleans, 55% were diagnosed among BMSM.

Along with higher rates of HIV infection than other groups, BMSM also face disparities in linkage and retention to HIV care, as well as in viral suppression. Among PWH, BMSM are less likely to be consistently engaged in HIV care than other groups with HIV and suffer from higher HIV-/AIDS-related mortality.² In 2014, 49% of BMSM were virally suppressed, compared to 63% of White MSM.³ Further, between 2017 and 2021, 70% of newly diagnosed AIDS cases were among Black people in Louisiana.⁴ In 2018, among those in HIV medical care, only 81% of BMSM aged 13–24 were virally suppressed compared to 98% of their White counterparts.

There is a growing body of literature that attributes lower rates of treatment adherence and viral suppression among BMSM to social and economic factors. BMSM are more likely to experience housing instability,⁵ trauma related to violence, incarceration, and sexual orientation-based discrimination,^6,7 and food insecurity compared to their White counterparts.^8,9 They are also more likely to experience medical mistrust because of racially and sexually based health care discrimination.¹⁰ Internalized shame and fear, because of pervasive community- and family-based stigma, has been linked to avoidance of HIV treatment adherence and persistent engagement in HIV care.² Thus, there are a number of social determinants of health (SDoH) and concomitant behavioral health conditions such as trauma, alcohol dependency, substance use, and depression that form barriers to antiretroviral therapy (ART) and HIV care adherence among BMSM.^7,11
–13

To address the needs of the BMSM population, we developed a novel program called +LOVE for BMSM PWH at CrescentCare, a Federally Qualified Health Center (FQHC), in New Orleans. CrescentCare has a 37-year history of providing HIV prevention and outreach services to residents of New Orleans and surrounding parishes and currently provides HIV care to 35% of individuals living with HIV (LWH) in the city. The agency also provides comprehensive primary care services as well as an array of support services. The +LOVE Program integrated nonmedical case management, the current support service standard in the HIV Care Continuum and at CrescentCare, with a multi-faceted and comprehensive integrated behavioral health crisis intervention designed to address both SDoH barriers as well as the behavioral health needs of our clients. We enrolled 146 BMSM PWH starting on October 7, 2019.

All participants were provided access to the interventions for 12 months, although some participants moved during the period or were not able to take advantage of all of the components of the intervention.

Four major intervention components comprised the +LOVE program: (1) the provision of a dedicated nonmedical case manager, who conducted screening assessments, provided referrals to resources, assisted with insurance and housing, and functioned as a liaison with the participants' HIV care team, following current Ryan White case management standards.¹⁴

In addition, clients were given access to (2) a dedicated behavioral health therapist, trained in trauma informed techniques, mindfulness based stress reduction, and cognitive behavioral therapy, who provided weekly 50 min therapy sessions; (3) an SMS (text message)-based engagement program, called CareSignal, which was tailored by the +LOVE team to administer ART medication reminders at times of day chosen by participants and delivered routine ecological momentary assessments (EMAs) to participants in the form of short, daily, weekly, or monthly text messages that inquired about the participants' basic needs and mental health status to identify barriers to HIV care engagement and treatment adherence. These were texts that queried whether participants had access to food, employment, and housing, for example, or comprised mental health assessments (including depression and anxiety tracking using modified Patient Health Questionnaire (PHQ)-9 and Generalized Anxiety Disorder (GAD)-7 scales^15,16 and questions related to mood and substance use).

The real-time information was utilized by the therapist to address participants' immediate needs and to tailor her therapy sessions and was used by the triage crisis coordinator; and (4) a triage crisis coordinator who responded to alerts generated by the EMA responses that met the threshold of crisis (e.g., high PHQ-9 scores, sudden mood changes, loss of housing) and who utilized the real-time information provided by CareSignal to assist participants' in handling crisis situations they were undergoing. The latter three intervention components, which comprise what we refer to as the “behavioral crisis intervention” consisted of support services that PWH at the agency did not have access to.

This study aims to assess the differential impact of the latter three components of the program (the dedicated therapist, the text message engagement program, and the triage crisis support), compared with the first component alone, a case manager, which is used as a proxy for the current standard of care, on access to HIV care, ART adherence, and viral suppression.

Methods

Recruitment and enrollment

We conducted recruitment for the +LOVE program from October 7, 2019 through December 31, 2021. Recruitment strategies consisted of (1) contacting current clients at our FQHC; (2) visiting venues and attending events at bars, clubs, and agencies where a large proportion of attendees were BMSM; (3) posting ads on dating apps popular with BMSM; (4) posting ads on bus stops and buses in neighborhoods and on routes where BMSM are known to reside, traverse, and/or visit; (5) sending digital flyers to other health facilities in the city; and (6) promoting peer to peer referrals by program enrollees. Inclusion criteria for participation were (1) confirmed HIV-positive serostatus; (2) 18 years of age and older; (3) identifying as BMSM; and (4) residing in New Orleans.

As the intervention represented an enhancement of current services and was considered a quality improvement project, Sterling IRB determined that it did not constitute human subjects research, and was, therefore, ineligible for IRB approval.

At enrollment, we introduced participants to the program and its various components. We made it clear that they could choose to take part in whichever component of the program that they wanted. All participants chose to enroll in the CareSignal text messaging app, although only 114 participants were active users; 56 participants accessed the integrated behavioral health/triage (BHT) crisis component; 51 participants accessed both the BHT/crisis component and case management; and 36 participants made use only of case management.

Most participants had worked with a case manager before enrollment, as the service is covered under Ryan White Part B funding for PWH whose income does not exceed 400% of the federal poverty level guidelines. Case management is the primary form of support services that PWH receive in the region. Participants were given the option to either remain with their current case manager or to work with the +LOVE program's dedicated case manager. The case manager provided similar support to participants as would be provided to nonprogram clients at CrescentCare. There were, however, key potential differences. The first was that because we had a case manager dedicated to the program, his client caseload was around half the size of the typical caseload at the agency.

Our case manager also had specific expertise working with BMSM in the community, he self-identified as a Black gay man, and he worked closely with the other staff members on the +LOVE team. Unlike other case managers, he also could provide participants with free Lyft rides to HIV care and related appointments, a component of the program that was funded by the grant. Because the case management provided by +LOVE represents an enhance version of basic case management at the agency, this component is only used as a proxy for basic services provided under Ryan White Part B.

Participants who chose the program case manager worked with him throughout their participation in the program. His interactions with participants were conducted through in-person appointments, phone calls, and text messaging directly to participants' phones. During the height of the COVID-19 public health emergency, most of his communication with participants was conducted remotely, although he did conduct in-person food delivery to participants.

An important part of the initial enrollment discussion with potential participants was psychoeducation, during which +LOVE staff discussed the BHT components of the program and addressed any stigma surrounding therapy and mental health care that was expressed by the participants. Prior CrescentCare data on client referrals and follow through for behavioral health services had revealed that therapy was relatively underutilized by Black clients compared to White clients. The +LOVE team was prepared to address stigma around mental health treatment, because of this background data and also in response to formative research that included community key informant interviews with BMSM revealed that stigma around behavioral health therapy and the use of psychotropic medication was quite prevalent among BMSM.

Psychoeducation was also incorporated into other aspects of the program, including case management and crisis management. Participants who chose to access the behavioral health therapist were provided with therapy appointments free of charge. The therapist had training and therapeutic expertise matched to the program population. She incorporated a social justice lens, was sex positive, and focused on building self-esteem. She was trained to work with clients who are not in traditional monogamous relationships. Among the therapeutic techniques she used were eye movement desensitization and reprocessing and mindfulness-based stress reduction.

The text-based messaging program provided various types of tracking and EMAs depending on individual participants' needs and comfort, to which they consented upon enrollment. Mood, depression, and anxiety tracking included periodic questions via texts asking about mood changes and questions that comprise the PHQ-9 scale for depression and the GAD-7 scale for anxiety,¹⁷ including questions regarding suicidal ideation and suicidality. Medication tracking provided reminders to take medications at times predesignated by the participants during enrollment and tracked adherence based on responses to the reminders. Basic needs tracking, which included predetermined SDoH indicators, inquired about participants' basic needs, such as food, housing, and employment.

Based on tracking parameters set by the +LOVE team, the text messaging service used text responses to classify participants according to their risk. For example, high numbers on mental health assessments, lack of response to medication alerts, and urgent basic needs triggered alerts that were sent directly to the therapist and the triage crisis coordinator, so that they could assess who needed immediate or more intensive attention.

The triage crisis coordinator, a licensed social worker, addressed and investigated all of the alerts that came through the text messaging system and determined what form of immediate linkage and support the client needed. She coordinated responses from the other team members as well. Basic needs and medication alerts, such as housing instability, inability to obtain medication, and food insecurity were dealt with by both the triage crisis coordinator and the case manager together. The triage coordinator conducted crisis management and wellness checks via phone calls and direct text messages to participants. The majority of alerts she responded to were related to behavioral health needs (37%). The second most common alerts were medication alerts (19%), in which clients required assistance with obtaining refills for their ART.

For participants in behavioral health crises whom she assessed as needing behavioral health therapy, she provided psychoeducation as a means of encouraging them to make an appointment with the program's dedicated therapist. The triage crisis coordinator was able to provide a different form of behavioral health support to participants, in particular to those who were less likely to engage in traditional therapy. Other alerts addressed by her related mainly to SDoH, such as food insecurity (14%), help with overdue utility bills (11%), and feeling “emotionally or physically unsafe where one lives” (11%). The latter included a wide range of concerns that participants had, none of which was an emergency.

The triage crisis coordinator addressed these with clients by providing crisis counseling and referrals to resources. For example, one client was concerned that his apartment was not up to current building codes, so the triage crisis coordinator assisted him with the process of using his Section 8 voucher to transfer to a new address. Another client expressed feeling unsafe in his current neighborhood, but owned his house and could not move. The triage crisis coordinator counseled him on efficient and affordable ways to better secure his home.

Data Collection Procedures and Outcomes Measures

We extracted electronic medical record data at baseline, 6-, and 12-month intervals, including information on participants' HIV care appointments, behavioral health therapy appointments, ART prescriptions, HIV-related laboratory appointments, and viral load levels. We conducted surveys with participants at baseline, 4-, 6-, 8-, and 12-month intervals, which covered a range of topics, including (1) demographics; (2) HIV care experiences; (3) self-reported adherence to HIV care; (4) use of ART; (5) attitudes toward behavioral health care; and (6) behavioral health assessments. We also extracted EMA response data from the text messaging program.

Analysis procedures were designed to assess the efficacy of an integrated behavioral health support model, including access to behavioral health therapy and engagement with the text-messaging service, CareSignal. Engagement in behavioral health therapy was measured as having attended a behavioral health therapy appointment during the intervention. Participation in the CareSignal messaging service was quantified by the number of messages sent by participants in response to messages received from CareSignal. These two program components were the primary predictors in all the statistical models that were constructed.

Outcomes of interest include (1) retention in HIV care, defined as having a routine HIV medical care visit in the last 6 months or as having two or more HIV visits within the past year; (2) ART adherence, defined in as having been prescribed ART in the last 12 months; and (3) viral suppression, defined as having an HIV viral load of <200 copies/mL at the last HIV viral load test in the last 12 months. Each outcome variable was assessed at baseline, 6-month follow-up, and 12-month follow-up. Measures from 12 months of follow-up serve as dependent variables in all models.

Utilization of case management services, defined as engaging with a case manager within the 12 months preceding the intervention or at any point within the intervention was included in three models. This variable definition was limited to a dichotomous variable, rather than a continuous measure corresponding to engagement in care, by the availability of data surrounding engagement with case management. Implications of this limitation will be further examined in the discussion section. By including case management and thereby controlling for beneficial effects from case management, the analysis more accurately simulates the benefits of an integrated intervention using behavioral health therapy and the support provided by CareSignal that exist beyond case management services, the current standard of care.

Age and PHQ-9 score were both assessed at baseline. Education, assessed at baseline was divided into two levels; those who had an education beyond high school and those who did not continue their education beyond high school. Unemployment was assessed at baseline; participants were considered to have a recent history of unemployment if they reported being unemployed at least three of the previous 12 months. Attitudes surrounding current substance use at baseline were assessed via degree of agreement with the statement “I feel like drugs or alcohol are currently causing issues in my life.” Responses were provided on a 7-point Likert scale ranging from strongly agree (1) to strongly disagree (7). Those who strongly disagreed were treated as the reference group.

Analysis

Frequencies stratified by engagement in behavioral health therapy and were calculated for all outcomes and covariates in the analysis. Frequencies are only reported by one primary predictor, behavioral health therapy, because engagement with CareSignal, one of the other primary predictors, was recorded on a continuous scale and treated as a continuous variable within the models. Fisher's exact test was conducted between behavioral health therapy engagement and all categorical covariates to assess baseline associations between treatment group and covariates (education, unemployment history, and attitude regarding current substance use). An analysis of variance (ANOVA) was used to assess baseline associations between behavioral health therapy engagement and continuous covariates (baseline age and baseline PHQ-9 score). Baseline associations between total number of responses sent through CareSignal and categorical covariates were assessed using ANOVA.

A test of correlation was conducted to assess baseline bivariate associations between total number of responses sent through CareSignal and continuous covariates. Confounding variables were selected into the model based on degree of association with program components of interest at baseline. Others were selected due to associations with the outcome of interest, and some were selected due to relevance in comparable studies. Missing data were found to be an issue in this cohort; 32 participants (22.54%) were found to have missing outcomes data and 5 participants (3.42%) had missing PHQ-9 score data. Four logistic regression models using multiple imputed data were constructed using SAS. Multiple imputation was conducted using Multiple Imputation by Chained Equations (MICE) methods.

Each model was constructed for a different outcome of interest (Tables 2 and 3). An effect modification term was included in models 1, 2, and 4 to assess potential statistical interactions between number of responses sent to CareSignal and engagement in behavioral health therapy.

Model 1 treated probability of a participant having had a routine HIV medical care visit in the past 6 months at 12 months follow-up. Primary predictors in the model included engagement in behavioral health therapy and total responses to CareSignal. The effect modification term was included in the model. A variable indicating engagement in case management was also integrated in the model to control for potential benefits to the participant generated by case management. A baseline measure of the outcome variable was included in the model to control for baseline magnitude of effect. Participant age at baseline, PHQ-9 score at baseline, education background, recent unemployment history, and attitudes surrounding current substance use at baseline were also included in the model.

Model 2 assessed probability of having had two or more HIV medical care visits in the past 12 months at 12 months follow-up. Engagement in behavioral health therapy, total responses to CareSignal, the effect modification term, engagement in case management, age at baseline, PHQ-9 score at baseline, education background, recent unemployment history, and attitudes surrounding current substance use at baseline were included in the model as covariates. Whether or not a participant had two or more HIV medical care visits in the past 12 months at baseline was included in the model to control for baseline magnitude of effect.

Model 3 utilizes probability of having been prescribed ART in the past 12 months as the dependent variable. Only engagement in behavioral health therapy and total number of responses were included in the model, however. Covariates were unable to be adequately fit into the model due to lack of variability in the outcome measure.

Whether a participant had a suppressed viral load of <200 copies/mL on the last viral load test within the past 12 months at 12-month follow-up was the dependent variable used in model 4. Engagement in behavioral health therapy, total number of CareSignal responses, the effect modification term, engagement in case management, and a measure of the outcome at baseline were all included in model 4. Other covariates selected into the model include age at baseline, PHQ score at baseline, education history, and recent unemployment history.

Results

Of the 146 participants in the program, 145 (99.32%) identified as Black or African American; the only participant that did not identify as Black, identified as another race not included in the listed options. Participants struggled with some SDoH; at baseline, 54.11% of participants reported binge drinking within the past 12 months, 59.59% of participants reported making $20,000 or less on a yearly basis, 10.27% of participants were incarcerated in the past 12 months, 55.17% reported being unemployed for at least three of the past 12 months, and 21.92% were experiencing housing instability.

The mean age of the study population was 34 [standard deviation (SD) = 9.97] (Table 1). Mean baseline PHQ-9 score was 6.59 (SD = 6.03), indicating mild depression (Table 1). Among participants in the study, 103 (70.55%) had achieved greater than a high school education (Table 1). At baseline, 25 participants (17.24%) somewhat strongly agreed that drugs or alcohol was a problem in their lives (Table 1).

Table 1.

Baseline Characteristics

	Received behavioral health therapy (n = 59)	Did not receive behavioral health therapy (n = 87)	Total (n = 146)
Age, mean (SD)	33.86 (8.30)	34.1 (11.05)	34 (9.97)
PHQ score, mean (SD)	7.91 (6.57)	5.69 (5.50)	6.59 (6.03)
Case management, n (%)
Received case management	39 (66.10)	50 (57.47)	89 (60.96)
Did not receive case management	20 (33.90)	37 (42.53)	57 (39.04)
Education, n (%)
Greater than high school	47 (79.66)	56 (64.37)	103 (70.55)
High school or less	12 (20.34)	31 (35.63)	42 (29.45)
Unemployment, n (%)
Unemployed in the past year	34 (57.63)	46 (53.49)	80 (55.17)
Employed throughout the past year	25 (42.37)	40 (46.51)	65 (44.83)
Substance use, n (%)
I feel like drugs or alcohol are currently causing issues in my life…
Strongly agree	3 (5.08)	6 (6.98)	9 (6.21)
Agree	2 (1.69)	3 (3.49)	4 (2.76)
Somewhat agree	6 (10.17)	6 (6.98)	12 (8.28)
Neither agree nor disagree	5 (8.47)	13 (15.12)	18 (12.41)
Somewhat disagree	1 (1.69)	2 (2.33)	3 (2.07)
Disagree	10 (16.95)	18 (20.93)	28 (19.31)
Strongly disagree	33 (55.93)	38 (44.19)	71 (48.97)

PHQ, Patient Health Questionnaire; SD, standard deviation.

Table 2.

Regression Coefficients and Standard Errors

Variable	Model 1		Model 2		Model 3		Model 4
Variable	Coefficient	SE	Coefficient	SE	Coefficient	SE	Coefficient	SE
Behavioral health therapy	0.8818^a	0.3362	1.2048^a	0.3863	−0.4913	0.5897	−0.4359	0.5259
CareSignal responses	0.0020	0.0011	0.0021	0.0013	−0.0023	0.0018	−0.0010	0.0014
Effect modifier term: behavioral health therapy and number of CareSignal responses	−0.0021	0.0011	−0.0016^a	0.0012	^*	^*	0.0020	0.0014
Case management	0.2053	0.2094	−0.2892	0.2335	^*	^*	0.2053	0.3927
Age at baseline	−0.3332	0.0187	0.0197	0.0201	^*	^*	0.0208	0.0417
PHQ score at baseline	0.0264	0.0346	0.0128	0.0376	^*	^*	−0.1016^a	0.0490
High school education or less	−0.0795	0.2158	−0.1407	0.2338	^*	^*	0.0876	0.3750
Unemployed 3 of last 12 months	−0.0812	0.2005	−0.2665	0.2171	^*	^*	−0.7104	0.3917
Substance use 1	0.3710	0.7267	−1.0354	0.8101	^*	^*	^*	^*
Substance use 2	0.1441	1.0221	0.0383	1.0742	^*	^*	^*	^*
Substance use 3	−0.4424	0.6540	−0.2381	0.7131	^*	^*	^*	^*
Substance use 4	−0.0508	0.5503	−0.2482	0.5872	^*	^*	^*	^*
Substance use 5	−0.7099	1.2288	0.5216	1.2840	^*	^*	^*	^*
Substance use 6	0.3274	0.4967	0.6360	0.5305	^*	^*	^*	^*
Did not have a routine HIV medical care visit in the past 6 months at baseline	−0.3695	0.2363	^*	^*	^*	^*	^*	^*
Did not have two or more HIV medical care visits in the past year	^*	^*	−0.5215	0.2502	^*	^*	^*	^*
Had HIV viral load <200 copies/mL at the last HIV viral load test in the last 12 months	^*	^*	^*	^*	^*	^*	1.3075	0.5123
Had HIV viral load ≥200 copies/mL at the last HIV viral load test in the last 12 months	^*	^*	^*	^*	^*	^*	−0.0612	0.5477

Statistical significance at the 0.05 level.

PHQ, Patient Health Questionnaire; SE, standard error.

Variable not included in model.

Table 3.

Odds Ratios and Associated p-Values

Variable	Model 1		Model 2		Model 3		Model 4
Variable	OR	p	OR	p	OR	p	OR	p
Behavioral health therapy	2.415^a	0.009	3.336^a	0.002	0.612	0.405	0.647	0.407
CareSignal responses	1.002	0.075	1.002	0.089	0.998	0.216	0.999	0.487
Effect modifier term: behavioral health therapy and number of CareSignal responses	0.998	0.053	0.998^a	0.001	^*	^*	1.002	0.150
Case management	0.998	0.112	0.749	0.216	^*	^*	1.228	0.601
Age at baseline	0.922	0.808	1.020	0.327	^*	^*	1.021	0.618
PHQ score at baseline	1.027	0.446	1.013	0.734	^*	^*	0.903^a	0.038
High school education or less	0.924	0.713	0.869	0.547	^*	^*	1.092	0.815
Unemployed 3 of last 12 months	0.922	0.686	0.766	0.220	^*	^*	0.491	0.070
Substance use 1	1.449	0.610	0.355	0.201	^*	^*	^*	^*
Substance use 2	1.155	0.888	1.039	0.972	^*	^*	^*	^*
Substance use 3	0.642	0.499	0.788	0.739	^*	^*	^*	^*
Substance use 4	0.950	0.926	0.780	0.673	^*	^*	^*	^*
Substance use 5	0.492	0.563	1.685	0.685	^*	^*	^*	^*
Substance use 6	1.387	0.510	1.889	0.231	^*	^*	^*	^*
Did not have a routine HIV medical care visit in the past 6 months at baseline	0.691	0.118	^*	^*	^*	^*	^*	^*
Did have two or more HIV medical care visits in the past year	^*	^*	0.594^a	0.037	^*	^*	^*	^*
Had HIV viral load <200 copies/mL at the last HIV viral load test in the last 12 months	^*	^*	^*	^*	^*	^*	3.697^a	0.011
Had HIV viral load ≥200 copies/mL at the last HIV viral load test in the last 12 months	^*	^*	^*	^*	^*	^*	0.941	0.911

Statistical significance at the 0.05 level.

OR, odds ratio; PHQ, Patient Health Questionnaire.

Variable not included in model, OR not generated.

Baseline PHQ-9 score was found to be significantly associated with engagement in behavioral health therapy at baseline (p = 0.0314). Via an ANOVA, total number of responses to CareSignal was found to be associated with recent unemployment history (p = 0.0265). All other covariates were found not to be significantly associated with the primary predictors at a 0.05 level.

Retention in care

Model 1

Those who engaged in behavioral health therapy were found to have 2.415 times higher odds of remaining engaged in HIV care, defined as having attended an HIV medical appointment within the past 6 months (Table 3). This value is statistically significant with a p-value of 0.009. For each response sent to a text message from CareSignal, the odds of having attended an HIV-related medical appointment in the past 6 months increased by 1.002 (Table 3). Considering the median number of responses sent to CareSignal is 201.5, this would indicate that the odds of having attended an HIV-related medical appointment in the past 6 months at baseline would be 1.496 times higher than those who do not engage in CareSignal at all. This effect is insignificant at the 0.05 level with a p-value of 0.0754 (Table 3).

Interestingly, there is a slight negative association between the effect modifier and the odds of having attended an HIV-related medical appointment in the past 6 months. If a participant is enrolled in behavioral health therapy, their odds of retention in care by this model's definition decrease by 0.998 for each response sent to CareSignal (Table 3). This interaction does not negate the effects of number of responses or behavioral health therapy, but it does modify the magnitude of effects. This interaction term is not significant at the 0.05 level as it has a p-value of 0.053 (Table 3). All potential confounders included in the model were found to be statistically insignificant.

Model 2

Those who engaged in behavioral health therapy were found to have 3.336 times higher odds of having attended two or more HIV-related medical visits in the past 12 months at 12 months follow-up than those who did not engage in behavioral health therapy (Table 3). This effect was statistically significant with a p-value of 0.002 (Table 3). The odds of having attended two or more HIV-related medical visits in the past 12 months increased by 1.002 for each text message response sent (Table 3). Considering this odds ratio, participants who meet the median number of text responses have 1.527 times higher odds of having had two HIV-related medical appointments than those who did not send any responses to CareSignal (Table 3).

This estimate, however, is not significant at the 0.05 significance level with a p-value of 0.0891 (Table 3). Regardless, the results suggest that increased engagement with CareSignal leads to an increased probability of achieving positive HIV-related health outcomes. A slight negative interaction between behavioral health therapy and text message responses exists; for each text response sent by someone who engaged with behavioral health therapy, the odds of having two HIV-related medical visits in the past 12 months decreased by 0.998 times (Table 3). This relationship is not significant at the 0.05 level and does not negate the positive effects of text message responses and behavioral health therapy.

Participants who reported that they did not have two HIV-related health care appointments in the past 12 months at baseline had 0.594 times lower odds of having two HIV-related health care appointments in the past 12 months at 12 months follow-up than those who did have two appointments at baseline. This baseline value was a significant confounding variable in the model (p = 0.0371).

ART adherence

Model 3

The model constructed to assess ART adherence at 12 months follow-up included the count of text message responses by each individual participant and engagement in behavioral health therapy. The model was fit using only behavioral health therapy and count of text responses as predictor variables due to lack of variability in the outcome variable. Text responses were found to have an insignificant negative relationship; each response was associated with 0.998 times the odds of being adherent to ART at 12 months follow-up (p = 0.216) (Table 3). Similarly, behavioral health therapy held a statistically insignificant negative relationship with ART adherence; engaging with behavioral health therapy was associated with 0.612 (p = 0.405) times the odds of being adherent to ART at 12 months follow-up (Table 3).

This negative relationship, however, should be contextualized within the lack of confounding variables included in the model, the lack of variability in the outcome measure, and the statistical retention of the null hypothesis. Potential benefits of additional research concerning the effects of similar integrative interventions and ART adherence will be further addressed in the discussion section of the article.

Viral suppression

Model 4

In the model assessing the association between participation in +Love program components and the probability of having a suppressed viral load at 12 months follow-up, both engagement in behavioral health therapy and number of responses to CareSignal messages were found to have a statistically insignificant negative relationship with the odds of having a suppressed viral load at 12 months follow-up. Each text response was associated with 0.999 times the odds of having a suppressed viral load at 12 months follow-up (p = 0.487) (Table 3). Participants who engaged in behavioral health therapy had 0.647 times the odds of having a suppressed viral load at 12 months follow-up than those who had not engaged in behavioral health therapy (p = 0.407) (Table 3).

The effect modification term to assess interaction between behavioral health therapies was found to have a positive association with increased odds of having a suppressed viral load at 12 months follow-up. Those who participated in behavioral health therapy held 1.002 times higher odds of having a suppressed viral load at 12 months follow-up; if a client participated in behavioral health therapy and sent the median 201.5 responses to CareSignal, they had 1.496 times higher odds of suppressed viral load at 12 months follow-up than those who did not participate in behavioral health therapy or send responses in CareSignal. This relationship, however, is not statistically significant with a p-value of 0.150 (Table 3).

Within the model, having a suppressed viral load at baseline and baseline PHQ score were both found to be significant confounders in the relationship between program components and viral suppression at 12 months. A suppressed viral load at baseline was associated with 3.697 times higher odds of a suppressed viral load at 12 months (p = 0.012) (Table 3). Each point increase in baseline PHQ score was found to be associated with 0.903 times lower odds of a suppressed viral load at 12 months (p = 0.038) (Table 3). Although not significant in this model at a 5% significance level, those who were reported to have been unemployed for three out of the past 12 months at baseline had 0.491 times lower odds of a suppressed viral load at 12 months follow-up (p = 0.070) (Table 3).

Discussion

The actual causes of HIV/AIDS disparities between BMSM and other groups are complex and involve interrelated social factors that are largely tied to the effects of historical and present-day institutionalized racism, including stigma and a lack of social support; impoverished neighborhoods; inequitable access to education and employment; inequities in incarceration rates; lack of comprehensive, adequate health care coverage due to affordability; lack of transportation to attend health care appointments; and homelessness. BMSM face heightened barriers around race, sexual orientation, HIV status, discrimination, and stigma when accessing appropriate medical care in the South. The local HIV epidemic intersects, then, with homophobia, racism, HIV stigma, and other key social and racial justice issues in New Orleans in many ways. Because of the complexity of factors, interventions aimed at improving HIV care outcomes for BMSM must address basic needs, social barriers, as well as the behavioral health needs of clients to be effective.^18,19

Currently, case management is the main form of support services available to BMSM LWH. While there is coverage available for behavioral health services, the reality is that many agencies that provide HIV care do not have such services in house, and there is a dearth of available therapists who can meet with clients in crisis. Even in health centers, such as CrescentCare that have in-house therapists, the wait is long to access appointments and the social barriers and stigma surrounding therapy prevent clients from following through on referrals.

Our findings support other research that has found mental health care to be an important facilitator in HIV care retention.²⁰ Our results suggest that an integrated behavioral health/crisis support model of intervention that makes use of a digital safety net, which can track SDoH indicators in real-time incorporated with case management, increases positive outcomes over case management alone. Implementation of such a model, however, must take into account that some groups among PLWH may have more hesitation than others in utilizing behavioral health care services and that the referral process may benefit from a psychoeducational component that helps to destigmatize these services.²¹ Furthermore, the types and forms of behavioral health services provided should be tailored to the desires and needs of the group. Traditional forms of in person, once a week therapy may not be the most efficacious means of providing behavioral health and/or crisis support.

The study does have its limitations, however. First, because it consisted of a convenience sample of people who chose to take part in the program, it lacks the extrapolative power of a randomized control trial. We are currently conducting a retrospective cohort analysis to assess HIV-related outcomes of the +LOVE participants and control patients who may provide a stronger case for the efficacy of the +LOVE program's behavioral health/support model. The control cohort is made up of patients seen by CrescentCare who met the inclusion criteria for the +LOVE program but did not participate in it.

Baseline differences between the treatment and control arms will be balanced using a propensity score algorithm that estimates the likelihood of participating in the +LOVE program given observable individual characteristics. Balance between the treatment and control arms will be evaluated using p-values and/or standardized mean differences on each baseline covariate. Outcomes will be compared between the postbalance treatment and control groups.

Second, as was described above, the case management provided in the +LOVE program represented an enhanced version of nonprogram case management.

Our analysis of ART adherence was greatly limited by an overall lack of variability in the ART adherence outcomes. The ability to select covariates such as case management and the baseline measure of the outcome variability limited the explorative scope of this model. A comparable study in a population with a lower degree of ART adherence may further enlighten the complexities of this relationship.

Finally, there is the issue of self-selection. Those people who engaged with the behavioral health/crisis components may be more inclined than others to seek help, to answer texts, and to want to engage in care. On the contrary, one could also imagine that if the +LOVE program's behavioral health/crisis support model became the standard of care, over time, through expanded psychoeducation and dissemination of outcomes to community, we could engage more people to make use of such services. In fact, in response to the positive outcomes of the +LOVE intervention, our agency has already extended the behavioral health/crisis components to other groups of PWH and groups who are at high risk for HIV.

Currently, an adaptation of these components is being utilized with transgender women at the agency to increase pre-exposure prophylaxis (PrEP) uptake and adherence, and there are plans to utilize the model among people who inject drugs to increase HCV medication adherence and treatment outcomes.

Notwithstanding these limitations, our findings show the potential benefits for those who engaged in behavioral health therapy and crisis support with the aid of text message SDoH and mental health tracking. In potential future applications of this program, we will continue to seek out those in need and engage them in care, thus eliminating the limiting effects of self-selection on program benefits. Despite these limitations, the results support the need for novel forms of behavioral health care and crisis support that are not necessarily represented in current standards of care for PWH.

Footnotes

Acknowledgments

We thank the following Special Programs of National Significance (SPNS) sites for their support and assistance: Sutter Bay Hospital—Easy Bay Advanced Care, Oakland, CA; Kimberly A Kisler, PhD, MPH, Friends Research Institute, Inc., California State University, Los Angeles Department of Public Health, Los Angeles, CA; Adan Cajina, MSc, Thelma Iheanyichukwu, MS, John Hannay, MPH, Chau Nguyen, MPH, Natalie Solomon-Brimage, MPH, Melinda Tinsley, MA and Joanne Hsu, MPH, HRSA, Rockville, MD; and representatives from Parkland Health and Hospital System, Dallas, TX; GMHC, New York, NY; Christian Community Health Center, Chicago, IL; Washington University, Project ARK, St. Louis, MO; Duke University Center for Health Policy & Inequalities Research, Durham, NC; and NORC at the University of Chicago, particularly Sarah Hodge, PhD, and David Rein, PhD.

Author Disclosure Statement

No competing financial interests exist.

Funding Information

This publication was made possible by the US Department of Health and Human Services, Health Resources and Services Administration under grant no. 18H97HA31809PL in the amount of US $975,000 awarded to New Orleans AIDS Task Force (dba CrescentCare) in New Orleans, LA. No percentage of this project was financed with nongovernmental sources.

References

Centers for Disease Control and Prevention (CDC). HIV/AIDS Surveillance Report, 2019. Division of HIV/AIDS Prevention, National Center for HIV/AIDS, Viral Hepatitis, STD, and TB Prevention, Centers for Disease Control and Prevention (CDC), US Department of Health and Human Services: Atlanta, GA, USA; 2021; 123 p.

Louisiana Office of Public Health STD/HIV Program. 2019. STD/HIV Surveillance Report. Department of Health, Office of Public Health: New Orleans, LA, USA; 2020; 122 p.

Louisiana Department of Health and Hospitals. PrIDE Executive Summary. Division of HIV/AIDS Prevention, National Center for HIV/AIDS, Viral Hepatitis, STD, and TB Prevention, Centers for Disease Control and Prevention (CDC), US Department of Health and Human Services: Atlanta, GA, USA; 2 p. Available from: https://www.cdc.gov/hiv/pdf/research/demonstration/executive_summaries/cdc-hiv-pride-Louisiana.pdf

Alexander

, Brown

, Bowen

, et al. Louisiana HIV/AIDS Strategy: For Prevention, Treatment and Care Services 2017–2021. Louisiana HIV Planning Group: New Orleans, LA, USA; 2016; 133 p.

Creasy

, Henderson

, Bukowski

, et al. HIV testing and ART adherence among unstably housed black men who have sex with men in the United States. AIDS Behav, 2019; 23(11):3044–3051; doi: 10.1007/s10461-019-02647-w

Harawa

, Brewer

, Buckman

, et al. HIV, sexually transmitted infection, and substance use continuum of care interventions among criminal justice-involved black men who have sex with men: A systematic review. Am J Public Health, 2018; 108(S4):e1–e9; doi: 10.2105/AJPH.2018.304698

Bogart

, Wagner

, Galvan

, et al. Longitudinal relationships between antiretroviral treatment adherence and discrimination due to HIV-serostatus, race, and sexual orientation among African–American men with HIV. Ann Behav Med, 2010; 40(2):184–190; doi: 10.1007/s12160-010-9200-x

Millett

, Peterson

, Flores

, et al. Comparisons of disparities and risks of HIV infection in black and other men who have sex with men in Canada, UK, and USA: A meta-analysis. Lancet, 2012; 380(9839):341–348; doi: 10.1016/S0140-6736(12)60899-X

Maulsby

, Millett

, Lindsey

, et al. HIV among black men who have sex with men (MSM) in the United States: A review of the literature. AIDS Behav, 2014; 18(1):10–25; doi: 10.1007/s10461-013-0476-2

10.

Louisiana Office of Public Health STD/HIV Program. 2018. STD/HIV Surveillance Report. Department of Health, Office of Public Health: New Orleans, LA, USA; 2019; 139 p.

11.

Harawa

, Williams

, Ramamurthi

, et al. Perceptions towards condom use, sexual activity, and HIV disclosure among HIV-positive African American men who have sex with men: Implications for heterosexual transmission. J Urban Health, 2006; 83(4):682–694; doi: 10.1007/s11524-006-9067-0

12.

Kraft

, Beeker

, Stokes

, et al. Finding the “community” in community-level HIV/AIDS interventions: Formative research with young African American men who have sex with men. Health Educ Behav, 2000; 27(4):430–441; doi: 10.1177/109019810002700406

13.

Wilson

, Moore

. Public health responses to the HIV epidemic among Black men who have sex with men: A qualitative study of health departments and 16 Understanding HIV Disparities in Louisiana communities in the US. AmJ Public Health, 2009; 99(6):1013–1022; doi: 10.2105/AJPH.2008.140681

14.

Louisiana Office of Public Health STD/HIV Program. Ryan White/HOPWA 2021 Service Definitions. Department of Health, Office of Public Health: New Orleans, LA, USA; 2020; 33 p.

15.

Kroenke

, Spitzer

, Williams

. The PHQ-9: Validity of a brief depression severity measure. J Gen Intern Med, 2001; 16(9):606–613; doi:10.1046/j.1525-1497.2001.016009606.x.

16.

Spitzer

, Kroenke

, Williams

, et al. A brief measure for assessing generalized anxiety disorder: The GAD-7. Arch Intern Med, 2006; 166(10):1092–1097; doi: 10.1001/archinte.166.10.1092

17.

US Health Resources & Services Administration. Patient Health Questionnaire (PHQ) Screeners. Available from: https://www.hrsa.gov/behavioral-health/patient-health-questionnaire-phq-screeners [Last accessed: September 06, 2022].

18.

Davoust

, Drainoni

, Baughman

, et al. He gave me spirit and hope: Client experiences with the implementation of community health worker programs in HIV care. AIDS Patient Care STDs, 2021; 35(8):318–326; doi: 10.1089/apc.2021.0085

19.

Dong

, Daudelin

, Koutoujian

, et al. Lessons learned from the pathways to community health study to evaluate the transition of care from jail to community for men with HIV. AIDS Patient Care STDs, 2021; 35(9):360–369; doi: 10.1089/apc.2021.0060

20.

Remien

, Stirratt

, Nguyen

, et al. Mental health and HIV/AIDS: The need for an integrated response. AIDS, 2019; 33(9):1411–1420; doi: 10.1097/QAD.000000000000227

21.

Hussen

, Camp

, Wondemeh

, et al. Mental health service utilization among young black gay, bisexual, and other men who have sex with men in HIV care: A retrospective cohort study. AIDS Patient Care STDs, 2021; 35(1):9–14; doi: 10.1089/apc.2020.0202