HIV Infections
Conditions
Brief summary
The proposed research will conduct a fully-powered efficacy trial of this approach in areas with large populations of AA and H/L MSM and high HIV incidence: Jackson, MS, Los Angeles, CA, and Boston, MA. High-risk MSM who have not tested for HIV in the last year will be recruited from MSM-oriented hook-up mobile apps, and assigned to receive either (1) HBST with post-test phone counseling/referral (eTEST condition), (2) standard HBST without active follow-up, or (3) reminders to get tested for HIV at a local clinic (control condition) at three month intervals over the course of 12 months. The investigators will explore the impact of the eTEST system on key outcomes, including rates of HIV testing, receipt of additional HIV prevention services, and PrEP initiation, compared with standard HBST or clinic-based testing reminders alone. The investigators will also explore the cost effectiveness of the eTEST system under various scenarios compared with relying on traditional, clinic-based testing alone.
Detailed description
HIV disproportionately affects men who have sex with men (MSM) in the United States, and new infections continue to increase particularly among African American (AA) and Hispanic/Latino (H/L) MSM. Past studies estimate that up to 50% of these new infections originate from the approximately 20% of MSM who are unaware of their status. Expanded HIV testing can produce reductions in incidence when implemented on a broad scale by facilitating earlier diagnosis and treatment. Rates of HIV testing are particularly low among AA and H/L MSM, and innovative approaches to encourage testing may help address high incidence in these men. Home-based, self-testing (HBST) for HIV offers considerable promise for increasing the number of MSM who are aware of their status by overcoming key barriers to clinic-based testing, such as inconvenience and confidentiality concerns. HBST may also be particularly well-suited for AA and H/L MSM, given that stigma and mistrust of medical care contribute to low testing rates. Despite its promise, however, many are concerned that HBST does not sufficiently connect users with critical post-testing resources, such as confirmatory testing and care among those who test positive, and that these limitations may result in delayed linkage to care. Existing, FDA-approved HBST kits provide a free, 24-hour helpline that offers these services to those who seek it, but few users do, and this passive approach may miss critical opportunities to engage with MSM for further prevention services. To address these challenges, the investigators developed a mobile health platform (eTEST) that uses internet-of-things (IoT) technologies to monitor when HBST users open their tests in real time, allowing the investigators to provide timely, active follow-up counseling and referral over the phone after they do so. In a pilot study, the investigators show that providing HBST by mail at regular intervals boosted rates of any/repeat HIV testing among high-risk MSM compared with clinic-based testing reminders. Moreover, those who received follow-up phone counseling after HBST were more likely to receive risk reduction counseling, to consult with a medical provider about PrEP, and to initiate PrEP. Given these promising results, the proposed research will conduct a fully-powered efficacy trial of this approach in areas with large populations of AA and H/L MSM and high HIV incidence: Jackson, MS, Los Angeles, CA, and Boston, MA. High-risk MSM who have not tested for HIV in the last year will be recruited from MSM-oriented hook-up mobile apps, and assigned to receive either (1) HBST with post-test phone counseling/referral (eTEST condition), (2) standard HBST without active follow-up, or (3) reminders to get tested for HIV at a local clinic (control condition) at three month intervals over the course of 12 months. The investigators will explore the impact of the eTEST system on key outcomes, including rates of HIV testing, receipt of additional HIV prevention services, and PrEP initiation, compared with standard HBST or clinic-based testing reminders alone. The investigators will also explore the cost effectiveness of the eTEST system under various scenarios compared with relying on traditional, clinic-based testing alone.
Interventions
DIAGNOSTIC_TESTHIV self-test
Home delivery of HIV self-test kits (OraSure OraQuick Rapid HIV test)
BEHAVIORALCounseling
Post-Test HIV Risk ReductionCounseling
Sponsors
National Institute of Mental Health (NIMH)
University of Southern California
The Miriam Hospital
University of Mississippi Medical Center
Brown University
Study design
Allocation
RANDOMIZED
Intervention model
PARALLEL
Primary purpose
PREVENTION
Masking
TRIPLE (Subject, Investigator, Outcomes Assessor)
Masking description
Participants are not informed of their condition assignment, but may infer it via the procedures they are provided. Both investigators and staff assessing outcomes are blinded to participants' group assignments.
Eligibility
Sex/Gender
MALE
Age
18 Years to No maximum
Healthy volunteers
Yes
Inclusion criteria
* report any of the following in the past six months: anal sex without condoms outside of a monogamous partnership with a recently tested, HIV-negative male, having been diagnosed with an STI, or being in an ongoing sexual partnership with an HIV-positive male * not tested for HIV in the last 12 months * have a stable residence in one of the site metros where they can securely receive packages * use an iOS/Android smartphone with a data plan or home wifi * fluent in either English or Spanish
Exclusion criteria
* currently on PrEP
Design outcomes
Primary
| Measure | Time frame | Description |
|---|---|---|
| Model Adjusted Probability of Any HIV Testing | 12 month study period | We used logistic regression with dummy-coded condition assignment as a predictor to test differences in outcomes across experimental conditions. A dummy-coded covariate indicating whether participants reported testing fewer than three times in the 3 years prior to enrolling was included in all models of HIV testing. We fit longitudinal mixed effects models for two outcomes, HIV testing and high-risk CAS events within a given follow-up period, given that these outcomes varied within participants across the study period. We specified distributions appropriate for each outcome (logistic for HIV testing and negative binomial for high-risk CAS events) with suitable link functions, unstructured covariance structures and robust standard errors. Time was included as a continuous covariate. A covariate reflecting pre-enrolment HIV testing and baseline CAS events were included in these models. We used an intent-to-treat approach for all analyses. Missing data were considered missing at random. |
| Model Adjusted Probabilities of Repeat HIV Testing (>1) | 12 months | We used logistic regression with dummy-coded condition assignment as a predictor to test differences in outcomes across experimental conditions. A dummy-coded covariate indicating whether participants reported testing fewer than three times in the 3 years prior to enrolling was included in all models of HIV testing. We fit longitudinal mixed effects models for two outcomes, HIV testing and high-risk CAS events within a given follow-up period, given that these outcomes varied within participants across the study period. We specified distributions appropriate for each outcome (logistic for HIV testing and negative binomial for high-risk CAS events) with suitable link functions, unstructured covariance structures and robust standard errors. Time was included as a continuous covariate. A covariate reflecting pre-enrolment HIV testing and baseline CAS events were included in these models. We used an intent-to-treat approach for all analyses. Missing data were considered missing at random. |
| HIV Diagnoses | 12 months | count of participants who were ultimately diagnosed with HIV during the course of the study |
Secondary
| Measure | Time frame | Description |
|---|---|---|
| Model Predicted Probability of Receipt of a Prescription for Pre-exposure Prophylaxis (PrEP) | 12 month study period | We used logistic regression with dummy-coded condition assignment as a predictor to test differences in outcomes across experimental conditions. A binary variable reflecting whether participants had ever had a PrEP prescription in the past was included for the PrEP prescription model. We specified two-way interactions between these covariates and condition assignment in all models, but none were significant and were excluded. |
| Model Predicted Probability of Receipt of Testing for Other Sexually-transmitted Infections | 12 months | We used logistic regression with dummy-coded condition assignment as a predictor to test differences in outcomes across experimental conditions. A dummy-coded covariate indicating whether participants reported testing fewer than three times in the 3 years prior to enrolling was included in all models of HIV testing. A similar covariate for STI testing was included in the STI testing model. We specified two-way interactions between these covariates and condition assignment in all models, but none were significant and were excluded. |
Other
| Measure | Time frame | Description |
|---|---|---|
| Average Predicted Number of High-risk Casual Anal Sex (CAS) Events With Partners of Unknown HIV and PrEP Status | 12 months | We used logistic regression with dummy-coded condition assignment as a predictor to test differences in outcomes across experimental conditions. We specified two-way interactions between these covariates and condition assignment in all models, but none were significant and were excluded. We fit longitudinal mixed effects models for two outcomes, HIV testing and high-risk CAS events within a given follow-up period, given that these outcomes varied within participants across the study period. We specified distributions appropriate for each outcome (logistic for HIV testing and negative binomial for high-risk CAS events) with suitable link functions, unstructured covariance structures and robust standard errors. Time was included as a continuous covariate. A covariate reflecting pre-enrolment HIV testing and baseline CAS events were included in these models. |
Countries
United States
Participant flow
Participants by arm
| Arm | Count |
|---|---|
| Control Participants will receive SMS text message reminders to get tested for HIV in a clinic. | 270 |
| Standard Self-Testing Participants will receive an HIV self-test kit in the mail with no standardized follow-up from counselors.
HIV self-test: Home delivery of HIV self-test kits (OraSure OraQuick Rapid HIV test) | 265 |
| Enhanced Self-Testing Participants will receive an HIV self-test kit and will be contacted via telephone for counseling within 24 hours of opening their test.
HIV self-test: Home delivery of HIV self-test kits (OraSure OraQuick Rapid HIV test)
Counseling: Post-Test HIV Risk ReductionCounseling | 275 |
| Total | 810 |
Withdrawals & dropouts
| Period | Reason | FG000 | FG001 | FG002 |
|---|---|---|---|---|
| Overall Study | Withdrawal by Subject | 1 | 0 | 0 |
Baseline characteristics
| Characteristic | Control | Total | Enhanced Self-Testing | Standard Self-Testing |
|---|---|---|---|---|
| Age, Continuous | 34.5 years STANDARD_DEVIATION 11.4 | 34.7 years STANDARD_DEVIATION 12.1 | 35.2 years STANDARD_DEVIATION 12.3 | 34.6 years STANDARD_DEVIATION 12.6 |
| Ethnicity (NIH/OMB) Hispanic or Latino | 95 Participants | 277 Participants | 98 Participants | 84 Participants |
| Ethnicity (NIH/OMB) Not Hispanic or Latino | 175 Participants | 533 Participants | 177 Participants | 181 Participants |
| Ethnicity (NIH/OMB) Unknown or Not Reported | 0 Participants | 0 Participants | 0 Participants | 0 Participants |
| Race (NIH/OMB) American Indian or Alaska Native | 2 Participants | 7 Participants | 3 Participants | 2 Participants |
| Race (NIH/OMB) Asian | 25 Participants | 72 Participants | 23 Participants | 24 Participants |
| Race (NIH/OMB) Black or African American | 21 Participants | 85 Participants | 31 Participants | 33 Participants |
| Race (NIH/OMB) More than one race | 20 Participants | 59 Participants | 23 Participants | 16 Participants |
| Race (NIH/OMB) Native Hawaiian or Other Pacific Islander | 0 Participants | 2 Participants | 1 Participants | 1 Participants |
| Race (NIH/OMB) Unknown or Not Reported | 23 Participants | 65 Participants | 22 Participants | 20 Participants |
| Race (NIH/OMB) White | 179 Participants | 520 Participants | 172 Participants | 169 Participants |
| Region of Enrollment United States | 270 participants | 810 participants | 275 participants | 265 participants |
| Sex/Gender, Customized Male | 269 Participants | 795 Participants | 268 Participants | 258 Participants |
| Sex/Gender, Customized Trans/Other gender identity | 1 Participants | 15 Participants | 7 Participants | 7 Participants |
Adverse events
| Event type | EG000 affected / at risk | EG001 affected / at risk | EG002 affected / at risk |
|---|---|---|---|
| deaths Total, all-cause mortality | 0 / 270 | 0 / 265 | 0 / 275 |
| other Total, other adverse events | 0 / 270 | 0 / 265 | 0 / 275 |
| serious Total, serious adverse events | 0 / 270 | 0 / 265 | 0 / 275 |
Outcome results
Primary
HIV Diagnoses
count of participants who were ultimately diagnosed with HIV during the course of the study
Time frame: 12 months
| Arm | Measure | Value (COUNT_OF_PARTICIPANTS) |
|---|---|---|
| Control | HIV Diagnoses | 0 Participants |
| Standard Self-Testing | HIV Diagnoses | 4 Participants |
| Enhanced Self-Testing | HIV Diagnoses | 4 Participants |
Primary
Model Adjusted Probabilities of Repeat HIV Testing (>1)
We used logistic regression with dummy-coded condition assignment as a predictor to test differences in outcomes across experimental conditions. A dummy-coded covariate indicating whether participants reported testing fewer than three times in the 3 years prior to enrolling was included in all models of HIV testing. We fit longitudinal mixed effects models for two outcomes, HIV testing and high-risk CAS events within a given follow-up period, given that these outcomes varied within participants across the study period. We specified distributions appropriate for each outcome (logistic for HIV testing and negative binomial for high-risk CAS events) with suitable link functions, unstructured covariance structures and robust standard errors. Time was included as a continuous covariate. A covariate reflecting pre-enrolment HIV testing and baseline CAS events were included in these models. We used an intent-to-treat approach for all analyses. Missing data were considered missing at random.
Time frame: 12 months
| Arm | Measure | Value (MEAN) |
|---|---|---|
| Control | Model Adjusted Probabilities of Repeat HIV Testing (>1) | 31.3 percent probability |
| Standard Self-Testing | Model Adjusted Probabilities of Repeat HIV Testing (>1) | 79.4 percent probability |
| Enhanced Self-Testing | Model Adjusted Probabilities of Repeat HIV Testing (>1) | 80.4 percent probability |
Primary
Model Adjusted Probability of Any HIV Testing
We used logistic regression with dummy-coded condition assignment as a predictor to test differences in outcomes across experimental conditions. A dummy-coded covariate indicating whether participants reported testing fewer than three times in the 3 years prior to enrolling was included in all models of HIV testing. We fit longitudinal mixed effects models for two outcomes, HIV testing and high-risk CAS events within a given follow-up period, given that these outcomes varied within participants across the study period. We specified distributions appropriate for each outcome (logistic for HIV testing and negative binomial for high-risk CAS events) with suitable link functions, unstructured covariance structures and robust standard errors. Time was included as a continuous covariate. A covariate reflecting pre-enrolment HIV testing and baseline CAS events were included in these models. We used an intent-to-treat approach for all analyses. Missing data were considered missing at random.
Time frame: 12 month study period
| Arm | Measure | Value (MEAN) |
|---|---|---|
| Control | Model Adjusted Probability of Any HIV Testing | 57.0 percent probability |
| Standard Self-Testing | Model Adjusted Probability of Any HIV Testing | 91.0 percent probability |
| Enhanced Self-Testing | Model Adjusted Probability of Any HIV Testing | 89.4 percent probability |
Secondary
Model Predicted Probability of Receipt of a Prescription for Pre-exposure Prophylaxis (PrEP)
We used logistic regression with dummy-coded condition assignment as a predictor to test differences in outcomes across experimental conditions. A binary variable reflecting whether participants had ever had a PrEP prescription in the past was included for the PrEP prescription model. We specified two-way interactions between these covariates and condition assignment in all models, but none were significant and were excluded.
Time frame: 12 month study period
| Arm | Measure | Value (MEAN) |
|---|---|---|
| Control | Model Predicted Probability of Receipt of a Prescription for Pre-exposure Prophylaxis (PrEP) | 20 percent probability |
| Standard Self-Testing | Model Predicted Probability of Receipt of a Prescription for Pre-exposure Prophylaxis (PrEP) | 15 percent probability |
| Enhanced Self-Testing | Model Predicted Probability of Receipt of a Prescription for Pre-exposure Prophylaxis (PrEP) | 15 percent probability |
Secondary
Model Predicted Probability of Receipt of Testing for Other Sexually-transmitted Infections
We used logistic regression with dummy-coded condition assignment as a predictor to test differences in outcomes across experimental conditions. A dummy-coded covariate indicating whether participants reported testing fewer than three times in the 3 years prior to enrolling was included in all models of HIV testing. A similar covariate for STI testing was included in the STI testing model. We specified two-way interactions between these covariates and condition assignment in all models, but none were significant and were excluded.
Time frame: 12 months
| Arm | Measure | Value (MEAN) |
|---|---|---|
| Control | Model Predicted Probability of Receipt of Testing for Other Sexually-transmitted Infections | 46 percent probability |
| Standard Self-Testing | Model Predicted Probability of Receipt of Testing for Other Sexually-transmitted Infections | 49 percent probability |
| Enhanced Self-Testing | Model Predicted Probability of Receipt of Testing for Other Sexually-transmitted Infections | 50 percent probability |
Other Pre-specified
Average Predicted Number of High-risk Casual Anal Sex (CAS) Events With Partners of Unknown HIV and PrEP Status
We used logistic regression with dummy-coded condition assignment as a predictor to test differences in outcomes across experimental conditions. We specified two-way interactions between these covariates and condition assignment in all models, but none were significant and were excluded. We fit longitudinal mixed effects models for two outcomes, HIV testing and high-risk CAS events within a given follow-up period, given that these outcomes varied within participants across the study period. We specified distributions appropriate for each outcome (logistic for HIV testing and negative binomial for high-risk CAS events) with suitable link functions, unstructured covariance structures and robust standard errors. Time was included as a continuous covariate. A covariate reflecting pre-enrolment HIV testing and baseline CAS events were included in these models.
Time frame: 12 months
| Arm | Measure | Value (MEAN) |
|---|---|---|
| Control | Average Predicted Number of High-risk Casual Anal Sex (CAS) Events With Partners of Unknown HIV and PrEP Status | 2.03 events |
| Standard Self-Testing | Average Predicted Number of High-risk Casual Anal Sex (CAS) Events With Partners of Unknown HIV and PrEP Status | 2.01 events |
| Enhanced Self-Testing | Average Predicted Number of High-risk Casual Anal Sex (CAS) Events With Partners of Unknown HIV and PrEP Status | 1.46 events |