Effect of case management on HIV outcomes for community corrections population: Results of an 18-month randomized controlled trial

Abstract

Background:

Evidence-based interventions that engage community-dwelling, justice-involved, people living with HIV (PLWH) in care are urgently needed. Project Bridge, an intensive case management intervention, has demonstrated efficacy for linking PLWH to care transitioning from prison to the community. We assessed whether a modified Project Bridge model was effective for increasing rates of HIV treatment engagement, antiretroviral therapy receipt and adherence for community-dwelling individuals supervised on probation and parole.

Setting:

Baltimore, Maryland

Methods:

In this study, the 18-month outcomes of a randomized controlled trial in which PLWH were also on probation or parole received either Project Bridge (n=50) or treatment as usual (n=50) were assessed. HIV treatment engagement (primary outcome), antiretroviral therapy prescription and adherence (secondary outcomes) are evaluated using the intent-to-treat approach.

Results:

There were no statistically significant differences in rates of HIV treatment engagement, antiretroviral therapy prescription receipt or adherence between groups over the 18-month study period. Across groups, participants were 5.6 times more likely to receive HIV care, 5.8 times more likely to receive an ART prescription, and 4 times more likely to report ART adherence at each follow-up period.

Conclusion:

Future research is needed to identify potentially less-intensive interventions that target the unique needs of PLWH under community supervision.

Introduction

Approximately 38,000 new human immunodeficiency virus (HIV) diagnoses occur in the U.S. annually¹, nearly 69% of which are transmitted by individuals who are not engaged in HIV medical treatment.² There is an urgent need to identify and disseminate evidence-based HIV prevention and treatment programs to reduce the spread of infection.

Ending the HIV Epidemic: A Plan for America, aims to reduce the incidence of HIV infections by 75% within five years.³ The initiative emphasizes greater access to HIV treatment for people living with HIV (PLWH) as a critical strategy for reducing HIV transmission.^3,4 Individuals who engage in HIV care and adhere to antiretroviral therapy (ART) may achieve viral suppression within six months of treatment initiation.⁵ People with an undetectable viral load (less than 200 copies of HIV per milliliter of blood) cannot transmit HIV to others.^6-8 However, efforts to increase the number of PLWH with viral suppression face two challenges: (1) identifying individuals who are infected but not engaged in medical care, and (2) effective strategies to mitigate barriers to treatment initiation and retention. These challenges substantially impact justice-involved PWLH, including individuals who are, or have, been incarcerated and/or are supervised by community corrections (i.e. probation, parole).

Health systems factors, logistical and social barriers may explain why community corrections supervised PLWH experience difficulties engaging in treatment. Multiple cycles of detention, release, and re-incarceration disrupt community-based HIV care engagement and ART adherence.⁹ Individuals re-entering the community may lack health insurance, and not know where or how to access HIV care.^10,11 Many individuals supervised by community corrections lack a usual source of health care,^9,11 which may impede regular HIV testing and delay treatment initiation. Additionally, transportation is a logistical barrier to accessing HIV care.^12,13 Community supervised individuals experience competing demands including supervision check-ins, mandated education/employment,¹⁴ and childcare¹⁵ which can limit individuals’ ability to attend medical appointments. Social barriers like stigma associated with visiting HIV-specific clinics,¹⁵ and taking ART in communal living spaces may disrupt adherence.^16,17

Research suggests patient navigators and case management may help mitigate such barriers.^14,18 However, prior interventions have focused on newly diagnosed PLWH in the general population or PLWH transitioning from incarceration to the community.^19-21 Little is known about the effect of case management services on HIV treatment engagement for community-dwelling PLWH on probation/parole.

This randomized controlled trial (RCT) aimed to determine whether Project Bridge (PB), an evidence-informed, community-based intensive case management strategy^19,22 was associated with greater rates of HIV (1) treatment engagement, (2) ART prescription and (3) adherence as compared to treatment as usual (TAU) for PWLH who are supervised by community corrections. We hypothesized that the PB would be associated with increased engagement, ART prescription and adherence compared to TAU.

Methods

This study responded to a national initiative aimed at engaging and retaining individuals with documented HIV infection who are “hard to reach” due to unstable contact with the healthcare system.²³ Baltimore, Maryland was an ideal study setting due to historically high rates of HIV infection and HIV-related risk behaviors like injection drug use.^24,25 When the study began, more than 10.5% of the city’s population was under community supervision.²⁶

Participant Recruitment

Participants were recruited from the Central Intake Unit (CIU) Probation and Parole Office between April 2011 and May 2015. Eligibility criteria included: (1) being 18 years or older, (2) under community supervision (probation/parole) at the time of recruitment, (3) residing in Baltimore, and (4) having documented HIV infection but not engaging in treatment. Individuals who had previously been in HIV care were eligible if they: (1) were unsatisfied with their provider, (2) did not attend a medical visit or receive a new ART prescription in the last 30 days, and (3) did not plan to re-engage in care with the past provider. We excluded individuals who did not meet eligibility criteria (e.g. those under pretrial supervision) or were unable to provide informed consent.

Our process for identifying PLWH on probation or parole has been described.^27,28 We conducted a randomized controlled trial to examine the acceptability of receiving rapid HIV testing at the CIU Probation and Parole Office. Participants were randomized to receive on-site testing or attend a community-based health clinic for testing.²⁷ Participants who tested positive at either site, or who self-identified as having a documented HIV infection were eligible for the second randomized controlled trial examining HIV care engagement and retention. To confirm study eligibility, we conducted rapid HIV testing for anyone who identified as having previously received an HIV diagnosis, and a confirmatory blood test for individuals who received an initial HIV diagnosis from the rapid testing study.

Randomization

Block randomization was conducted so that for each block of four participants, an equal number of participants were assigned, within gender, to PB and TAU. Neither study staff nor participants could be blinded to group assignment due to the nature of the intervention. Research assistants administered baseline assessments prior to randomization to minimize measurement bias.

Following randomization, TAU participants received a list of HIV clinics in Baltimore. If TAU participants did not schedule or attend a medical appointment to have their viral load measured within six months post-randomization, they would be offered a “rescue” opportunity to crossover to the PB intervention group. Participants randomized to PB were scheduled for an initial appointment at a community health center in Baltimore where staff were prepared to provide the PB model.

Project Bridge Intensive Case Management Intervention

PB model was initially developed by The Miriam Hospital in Providence, Rhode Island to promote continuity of medical care for PLWH transitioning from incarceration to the community. In the original model, PB patients receive co-located medical care and case management services provided by a professional social worker and a paraprofessional assistant. PB providers conduct a psychosocial needs assessment, coordinate medical and behavioral healthcare, serve as medical liaisons between physicians and the patient, and advocate for patients’ medical needs or entitlements. Prior research suggests PB is associated with increased healthcare retention for individuals transitioning from incarceration to the community.^19,22

Our study adapted the PB model to serve community-supervised individuals who were not necessarily transitioning out of incarceration. Medical doctors, nurses, social workers involved in the original PB intervention were consulted on PB adaptations for this study including services to support substance use treatment, housing, employment and legal needs. PB adaptations did not include feedback from justice-involved PWLH. Researchers then worked with an established community clinic to deliver PB services. The PB case management team included a licensed clinical social worker and an outreach worker who conducted psychosocial assessments with participants to develop individualized treatment plans. Participants were asked to meet with case managers weekly for the first three months to stabilize treatment engagement. Participants received PB services for 12-months. Researchers monitored the case management team’s fidelity to PB for all participants using a monthly checklist to track service delivery and outreach efforts (Appendix 1).

Data Collection

Participants completed baseline, 3-, 6-, 12-, 15- and 18-month follow-up assessments. Baseline assessments were conducted in a private space within the city’s probation/parole CIU. Follow-up assessments were conducted in confidential settings at the research team’s office, CIU, or in a jail/prison if the participant was incarcerated during the follow-up period. All assessments were administered by trained research staff. Participants were compensated $20 for the baseline assessment and $30 for each follow-up assessment.

Baseline and follow-up assessments inquired about participant demographics, general health status²⁹, risky sexual (e.g. unprotected sex) and substance use (e.g. injection) behaviors³⁰, risk for depression³¹, lifetime and recent crimes and criminal justice involvement.^32,33 Follow-up assessments also asked about participants’ perceived access to and receipt of medical care³⁴, fART adherence³⁵, and social supports.³⁶ Medical record review was conducted to verify self-reported engagement in medical care and assess changes in participant viral loads over time. However, inconsistent receipt of medical records and viral load data for control group participants prohibited analysis of viral load changes.

Study Outcomes

Treatment engagement, the primary outcome, was defined as an individual having received care for HIV/AIDS in the previous 90 days (yes, no). The secondary outcome of treatment retention was measured by: (1) receipt of ART prescription (yes, no); and (2) ART adherence, measured as the proportion of ART consumed in the previous 30 days. Initial inspection of ART adherence data revealed a bimodal distribution with responses clustered around 0% and 100% adherence. To address this issue, ART adherence was treated as a binary variable such that patient reported adherence ≤95% was treated as ‘non-adherent’, and >95% represented ‘adherent’. These cut-off points correspond with prior research indicating that very high levels of ART adherence (>95%) are required to achieve virologic suppression.³⁷

Statistical Analysis

We employed an intent-to-treat approach whereby all participants were analyzed according their randomization group. We conducted sensitivity analyses on all primary outcomes using an as treated approach to account for participants who received the “rescue” option to crossover from TAU to PB. Descriptive statistics were calculated to characterize the study sample and assess potential differences in the treatment and control groups at baseline.

HIV treatment engagement, ART receipt and adherence were evaluated using generalized linear mixed models (GLMM) to analyze the effect of treatment condition (i.e. PB vs. TAU) while controlling for the effects of assessment time point. Data gathered at follow-up periods were treated as observations nested within individuals and analyzed hierarchically. Logistic regression was used to analyze primary outcomes. All analyses employed conditional growth models where treatment assignment was a predictor of the intercept (main effect of treatment condition) and the slope of time (interaction effect between treatment condition and time). We considered homelessness (yes vs. no) and insurance status (insured vs. uninsured) as potential time-varying covariates and moderators of the effect of treatment condition on each outcome variable. To circumvent issues with model convergence, random-effect terms were specified as uncorrelated.³⁸ Terms were deleted from each model if they were found to be non-significant. Statistical tests were performed using R 3.6.2 and the lme4 package.³⁹

Ethics and Confidentiality Protections

This research was approved by Friends Research Institute’s Institutional Review Board and the Maryland Department of Public Safety and Correctional Services research committee. Researchers obtained a Federal Certificate of Confidentiality to protect participants. Research staff obtained informed consent from all study participants after reviewing the study intent, procedures, potential risks, benefits and confidentiality. Participation in this study did not impact participants’ current or future supervision status. This study was registered at ClinicalTrials.gov (NCT01366495).

Results

Participant Characteristics

Of the 240 individuals screened, 100 eligible individuals were randomized. The primary reason for study ineligibility was that individuals were already engaged in HIV care. Table 1 presents baseline characteristics for the 100 randomized participants. PB and TAU treatment groups were balanced across demographic variables. On average, participants were 45.2 years old (SD = 7.8). Most participants were Black (96%) men (78%). Only 3% of participants were employed, 74.7% lived below the poverty level. Thirty-four percent of participants were uninsured at baseline, and approximately half were either homeless (38%) or had an unstable housing status (14%). On average, participants had been living with HIV for 13.9 years (SD = 7.8). Most participants (69%) were supervised on probation, experienced an average of 11 incarceration episodes and nearly 8 years of jail or prison time.

Table 1.

Summary of Demographic Characteristics at Baseline

	Project Bridge (PB; N=50)	Treatment as Usual (TAU; N=50)	Total Sample (N=100)
Age (years), M (SD)	46.6 (6.7)	43.9 (8.6)	45.2 (7.8)
Sex, n (%)
Male	39 (78)	39 (78)	78 (78)
Female	11 (22)	11 (22)	22 (22)
Race1 n (%)
Black	48 (96)	48 (96)	96 (96)
White	1 (2)	2 (4)	3 (3)
Hispanic	2 (4)	1 (2)	3 (3)
Native American	1 (2)	2 (4)	3 (3)
Asian	1 (2)	0 (0)	1 (1)
Other	0 (0)	1 (2)	1 (1)
Education, n (%)
Less than high school	18 (36)	16 (32)	34 (34)
High school graduate or GED	20 (40)	20 (40)	40 (40)
College or higher	11 (22)	14 (28)	25 (25)
Missing	1 (2)	0 (0)	1 (1)
Employed, n (%)	1 (2)	2 (4)	3 (3)
Marital status, n (%)
Married	3 (6)	4 (8)	7 (7)
Single	18 (36)	30 (60)	48 (48)
Divorced	5 (10)	1 (2)	6 (6)
Widowed	4 (8)	0	4 (4)
Separated	6 (12)	4 (8)	10 (10)
Living with partner	14 (28)	11 (22)	25 (25)
Income (USD), M (SD)	$13,155.98 ($16,764.79)	$9,348.85 ($12,559.69)	$11,231.49 ($14,830.46)
Below poverty level2, n (%)	33 (73.3)	35 (76.1)	68 (74.7)
Health Insurance, n (%)
Medicaid	9 (18)	12 (24)	21 (21)
Medicare	7 (14)	6 (12)	13 (13)
Private	0 (0)	1 (2)	1 (1)
Uninsured	18 (36)	16 (32)	34 (34)
Homeless, n (%)
Yes	16 (32)	22 (44)	38 (38)
Don’t Know/ Unstable	7 (14)	7 (14)	14 (14)
Years living with HIV, M (SD)	14.3 (7.4)	13.5 (8.3)	13.9 (7.8)
Individuals newly diagnosed with HIV in the prior testing RCT, n (%)	1 (2)	2 (4)	3 (3)
Number of days of substance use during the past 90 days, M (SD)	17.8 (30.3)	20.2 (32.8)	19.0 (31.4)
Alcohol	13.4 (28.0)	15.3 (28.7)	14.3 (28.2)
Cocaine	0.1 (0.3)	0.8 (4.5)	0.4 (3.2)
Heroin	5.7 (18.5)	6.0 (21.6)	5.8 (20.0)
Injection drug use	5.5 (21.0)	3.9 (17.8)	4.7 (19.4)
Number of substance use treatment episodes over lifetime, M (SD)	3.1 (3.3)	3.1 (3.1)	3.1 (3.2)
Ever received buprenorphine treatment (Lifetime), n (%)	17 (34)	13 (26)	30 (30)
Ever received methadone maintenance, n (%)	13 (26)	15 (30)	28 (28)
At risk for depression	36 (72)	32 (64)	68 (68)
Supervision status, n (%)
Probation	33 (66)	36 (72)	69 (69)
Parole	13 (26)	12 (24)	25 (25)
Both	4 (8)	2 (4)	6 (6)
Days under current community supervision status, M (SD)	289.5 (429.9)	269.3 (352.2)	279.6 (391.9)
Days detained or incarcerated in the past 90 days, M (SD)	15.0 (29.2)	21.2 (32.2)	18.1 (30.7)
Age at first arrest (years), M (SD)	18.3 (6.4)	18.2 (5.1)	18.3 (5.7)
Lifetime arrests, M (SD)	16.7 (15.4)	14.8 (11.2)	15.7 (13.4)
Age at first incarceration, M (SD)	20.4 (6.6)	20.80 (7.6)	20.6 (7.1)
Lifetime number of incarcerations, M (SD)	11.4 (10.5)	10.6 (8.8)	11.0 (9.62)
Lifetime months incarcerated, M (SD)	101.2 (97.0)	95.2 (83.8)	98.2 (90.2)

Notes:

¹Participants could identify with multiple racial groups.

²Based on a household size of 1 (more conservative).

Follow-up rates for the PB intervention and TAU groups are summarized in Figure 1. Five participants received the “rescue” option to cross-over to PB. Nine participants (PB, n= 3; TAU, n= 6) became unreachable, despite researchers’ attempts to make contact, and six participants died (PB, n=3; TAU n=3).

Figure 1. Study Consolidated Standards for Reporting Trials (CONSORT) Flow Diagram.

Notes: After the 3-month follow-up period, three TAU participants had not engaged in HIV care and were therefore “rescued” into the PB group. After the 6-month follow-up period, two additional TAU participants were “rescued” into PB after dropping out of TAU care. Throughout the study period, six participants died (PB = 3, TAU = 3) and nine were lost to follow-up (PB=3, TAU = 6).

HIV Treatment Engagement

At baseline, 26.5% of PB and 35.4% of TAU participants reported having received HIV care in the past 90 days, although none had scheduled HIV-related medical appointments or planned to return to past providers for services or to renew prescriptions (consistent with eligibility criteria; Table 2). By 18-month follow-up, 93.2% and 92.7% of PB and TAU participants, respectively, had engaged in HIV treatment as indicated by self-reported receipt of HIV medical care during the last 90 days (verified by medical record review when possible). The GLMM model showed that assessment time point had a significant effect on treatment engagement when controlling for the effects of treatment condition (b = 1.72, SE = 0.36, z = 4.81, p < 0.001). Across treatment conditions, participants were 5.6 times more likely to report having received HIV care than at the previous assessment. Controlling for time revealed a significant effect of treatment condition on the likelihood of receiving HIV care (b = −2.09, SE = 0.93, z = −2.25, p = 0.02) favoring TAU over PB, but no significant main effect of insurance on the likelihood of receiving HIV care (b = 0.32, SE = 0.63, z = 0.50, p = 0.61). This result was qualified by a significant interaction effect between treatment condition and insurance status (b = 2.26, SE = 1.00, z = 2.26, p = 0.02). To probe this effect at differing levels of the interaction terms, simple slope tests were performed [47,48]. Amongst all uninsured participants, those assigned to TAU were 10.8 times more likely to receive HIV care at any given assessment time point. There were no significant differences amongst all insured participants, or between the insured and uninsured TAU participants. Amongst PB participants, insured individuals were 18.7 times more likely to receive HIV care at a given follow-up period compared to uninsured individuals. No significant interaction effect between time and treatment condition was identified [b = 0.05, SE = 0.35, z = 0.13, p = 0.90].

Table 2.

Summary of HIV treatment-related outcomes over the 18-month follow-up period

Outcome	Condition	Baseline	3-Month	6-Month	12-Month	15-Month	18-Month
Receipt of HIV medical treatment in the past 90 days, n (%)	Project Bridge	13 (26.5)	33 (67.3)	38 (80.9)	42 (89.4)	42 (93.3)	41 (93.2)
	Treatment as Usual	17 (35.4)	35 (70)	37 (75.5)	40 (93.0)	40 (93.0)	38 (92.7)
Prescribed antiretroviral therapy, n (%)	Project Bridge	22 (44.9)	23 (46.9)	32 (68.1)	37 (78.7)	40 (88.9)	36 (87.8)
	Treatment as Usual	20 (41.7)	31 (62)	38 (77.6)	36 (85.7)	38 (88.4)	38 (86.4)
Adherent to antiretroviral therapy medications in the past 30 days (>95% taken), n (%)	Project Bridge	12 (24.5)	16 (32.7)	32 (68.1)	34 (72.3)	31 (68.9)	35 (79.5)
	Treatment as Usual	8 (16.7)	26 (52)	29 (59.2)	32 (74.4)	34 (79.1)	33 (80.5)

The main effect of treatment condition and the interaction effect between treatment condition and insurance status did not remain significant in sensitivity analyses using an as treated approach (treatment condition: [b = −1.04, SE = 0.86, z = −0.54, p = 0.23]; condition x insurance: [b = 1.45, SE = 0.94, z = 1.54, p = 0.12]. The significant effect of time point was not significantly altered [b = 1.82, SE = 0.37, z = 4.88, p < 0.001].

Receipt of ART Prescription

Almost half of the sample were prescribed ART at baseline (44.9% PB, 41.7% TAU). At the 18-month follow-up, 87.8% of PB and 86.4% of TAU participants were prescribed ART. The increase in ART prescriptions over time was statistically significant (b = 1.75, SE = 0.35, z = 5.07, p < 0.001), as all participants were 5.8 times more likely to receive a prescription for ART than at the previous assessment. However, when controlling for this effect there were no significant differences in the likelihood of having received ART between treatment conditions (b = −0.11, SE = 0.71, z = −0.16, p = 0.87), or between the two conditions over time (b = −0.31, SE = 0.36, z = −0.87, p = 0.39). Neither homelessness nor insurance status produced significant improvements as time varying covariates and moderators of treatment condition, and were deleted from the model. An as treated approach did not alter the results with regards to the receipt of ART prescriptions.

ART Medication Adherence

Amongst those prescribed ART, 24.5% and 16.7% of PB and TAU participants respectively, adhered to their medication regimens (>95% adherence rate) at baseline. After 18 months, rates of ART adherence increased to 79.5% for PB and 80.5% for TAU. Across treatment conditions, the increase in 30-day ART adherence over time was statistically significant (b = 1.39, SE = 0.26, z = 5.36, p < 0.001). From one assessment time point to the next, all study participants were 4 times more likely to report ART adherence (>95% of medication taken). There was no statistically significant effect of group assignment on self-reported ART adherence (b = 0.004, SE = 0.54, z = 0.01, p = 0.99), nor was there a significant effect of group assignment over time (b = −0.23, SE = 0.29, z = −0.79, p = 0.43). An as treated approach did not alter the results with regards to ART medication adherence.

Discussion

The intensive case management PB intervention was no more effective than TAU in engaging community supervised PLWH in HIV care. There were no statistically significant differences in rates of ART prescription receipt or ART adherence between PB and TAU participants. Across treatment conditions, participants achieved significant increases in rates of HIV care, ART prescription and ART adherence overtime.

Treatment engagement is the second step in the HIV care continuum and is an imperative prerequisite for obtaining ART and achieving viral suppression. Low rates of HIV care measured at baseline reinforce prior documentation of widespread HIV treatment disengagement amongst PLWH who are supervised by community corrections PLWH.^9,21,42,43 Our results provide evidence that HIV treatment engagement is a persistent challenge for community supervised PLWH.

In our study, rates of insurance did not differ between treatment groups, and insured TAU were more likely than uninsured PB participants to receive HIV care. This suggests that community-based HIV treatment engagement interventions must address structural barriers (e.g. insurance) as well as social determinants of health (e.g. homelessness, unemployment) in order to increase treatment engagement.^44,45 Prior justice-setting based interventions that focused on removing healthcare access and financial barriers for incarcerated individuals and those transitioning to community re-entry have achieved increased rates of HIV treatment engagement.^19,22,46 This is the first study to examine the effectiveness of a community-based HIV treatment engagement initiative targeting community supervised PLWH. We found that the PB model proven successful for individuals during the re-entry transition may not be appropriate for individuals who are already established in the community. The limited impact of our community-based PB intervention may be explained by (1) community-based, justice-involved PLWH requiring a less-intensive prompt to stimulate HIV care utilization, and (2) undocumented commonalities between the PB and TAU care models.

Improvements in engagement, ART receipt and adherence across treatment groups suggests that the intensive PB model may not be necessary to improve outcomes. Instead, a less intensive prompt may be sufficient to promote treatment entry and initiate the HIV care cascade.⁴³ Similarly, Wohl et al. found that a case management intervention was no more effective than TAU for linking justice-involved PLWH to HIV care immediately following release from prison.²¹ In our study, participants had been living with HIV for almost 14 years. It’s possible that the confirmatory HIV testing required for study participation and repeated follow-up periods when researchers inquired about HIV care were sufficient to prompt engagement and ART adherence, thus creating a Hawthorne effect.⁴⁷ To evaluate this effect, future research should investigate HIV treatment engagement rates over a longer follow-up period with fewer intermittent assessments. Qualitative interviews with community supervised PLWH could reveal which barriers and facilitators to treatment engagement and ART adherence are most impactful and reveal how to tailor less-intensive interventions to community supervised PLWH.

Undocumented commonalities between PB and TAU care models, and overall access to HIV care in Baltimore may also explain outcome improvements across both treatment conditions. Core components of PB included a case management team, transportation to medical appointments, and enrollment in social services.¹⁹ Some of these PB components may have been provided by TAU providers. Baltimore’s status as a geographic hotspot for HIV transmissions prompted the city’s public and private organizations to expand access to both HIV care and related social services including integrated medical and psychosocial care and patient navigation services for PLWH from low-income populations.⁴⁸ Such services may have similar engagement benefits as compared to our PB model. While TAU providers would not have specifically sought out justice-involved PLWH, almost 75% of study participants were living below the federal poverty level and would qualify for such services. Additionally, of the 240 individuals screened for this study, 132 were ineligible because they were already engaged in care, suggesting moderate access to care exists in Baltimore.

Improved HIV outcomes across treatment groups highlights the need to examine each PB component separately, as opposed to measuring impact from the bundled model of care. One approach could utilize adaptive treatment strategies or stepped-care interventions where certain aspects of care are intensified or softened depending on patient responsiveness.^49,50 Such research would reveal which components have the greatest impact on HIV outcomes, and at which intensity levels.

Strengths and Limitations

This study achieved robust follow-up rates overtime and collected data on three aspects of the HIV care cascade: treatment engagement, ART prescription and medication adherence. Treatment engagement was confirmed by reviewing participant medical records at each follow-up. It was not possible to obtain reliable viral load data for all of the TAU participants in this study, which prevented analyzing clinical differences across groups. All participants signed releases of information for their medical records, and the researchers made several attempts to obtain records from TAU clinics. Inconsistent receipt of TAU clinical records across follow-up periods produced significant missingness in the viral load outcome, and were therefore excluded from the analysis. ART adherence was self-reported and may be vulnerable to response bias or social desirability bias. Finally, these results may not be generalizable to other communities that differ in the accessibility or comprehensiveness of HIV services.

Conclusion

Project Bridge, an intensive case management intervention was no more effective than TAU s in increasing HIV treatment engagement, ART prescription and adherence rates for community supervised PLWH. Future research is needed to identify potentially less-intensive interventions that target the unique needs of PLWH under community supervision.