Behavioral Economics Incentives to Support HIV Treatment Adherence (BEST) in Uganda
Last registered on January 25, 2019

Pre-Trial

Trial Information
General Information
Title
Behavioral Economics Incentives to Support HIV Treatment Adherence (BEST) in Uganda
RCT ID
AEARCTR-0003791
Initial registration date
January 24, 2019
Last updated
January 25, 2019 3:57 AM EST
Location(s)

This section is unavailable to the public. Use the button below to request access to this information.

Request Information
Primary Investigator
Affiliation
RAND Corporation
Other Primary Investigator(s)
Additional Trial Information
Status
On going
Start date
2018-04-12
End date
2023-03-01
Secondary IDs
NCT03494777
Abstract
Background: Many HIV-positive patients do not adhere to their antiretroviral medication (ART). This leads to higher viral loads and greater probability of transmission. Present bias—a tendency to give in to short-term temptations at the expense of long-term outcomes—is a potential driver of low adherence. In this study we test a novel intervention rooted in behavioral economics that is designed to overcome present bias and increase ART adherence.

Methods/Design: We will enroll 400 HIV-positive patients (330 core sample, and 70 additional as described below) at Mildmay hospital in Kampala, Uganda in a two-year randomized controlled trial. Participants will be randomized to one of three groups. The first intervention group (T1, n=110) will be eligible for small lottery prizes based on timely clinic visits and demonstration of viral suppression. Group 2 (T2, n=110) will be eligible for the same lottery prizes conditional on high electronically measured adherence as captured by a medication event management system (MEMS)-cap. The control group (n=110) will receive the usual standard of care. In addition, 70 treatment initiators (those who started treatment within the last three months) will also be recruited and assigned to T2. Adherence will be measured continuously throughout the study and for 12 months post-intervention to evaluate effect persistence, surveys will be conducted at baseline and then every six months, and viral loads will be measured annually. Primary outcomes are undetectable viral load and MEMS-measured adherence. Secondary outcomes are log transformed viral load as a continuous measure and a binary measure for whether the person adhered to greater than 90 percent of their ART pills.

Discussion: Our study is one of the first to investigate the effectiveness of lottery incentives for improving ART adherence, and in addition compares the relative efficacy of using measured adherence versus timely clinic visits and suppressed viral load to determine lottery eligibility. MEMS-caps are relatively costly, whereas viral load testing is now part of routine clinical care in Uganda. BEST will test whether directly incentivizing timely clinic visits and viral suppression (which canbe implemented using readily available clinic data) is as effective as incentivizing electronically measured adherence. Cost-effectiveness analyses of the two implementation modes will also be performed.
External Link(s)
Registration Citation
Citation
Linnemayr, Sebastian. 2019. "Behavioral Economics Incentives to Support HIV Treatment Adherence (BEST) in Uganda." AEA RCT Registry. January 25. https://www.socialscienceregistry.org/trials/3791/history/40640
Sponsors & Partners

There are documents in this trial unavailable to the public. Use the button below to request access to this information.

Request Information
Experimental Details
Interventions
Intervention(s)
There will be two intervention arms, both of which will use lottery-based incentives. In Treatment Group 1 (T1), clients will be eligible for quarter-yearly lotteries with small prizes based on timely drug refills and annual large prize lotteries if they demonstrate viral suppression. In Treatment Group 2 (T2), clients will be eligibility for quarterly small prize lotteries and annual larger prize lotteries based on high adherence as measured by electronic devices known as medication event management systems (or MEMS-caps). The control group will receive care as usual, including any adherence support mechanisms that are present within usual care practices. All participants including those in the control group will receive MEMS-caps.
Intervention Start Date
2018-04-12
Intervention End Date
2022-03-01
Primary Outcomes
Primary Outcomes (end points)
We will have one biological primary outcome and one behavioral primary outcome, namely viral suppression, and the percent of prescribed medication taken (mean adherence), respectively.

Primary Outcomes (explanation)
Our behavioral primary outcome will be electronically-monitored adherence. MEMS-data will be collected continuously over the course of the 24-month study period allowing us to investigate daily adherence and its timing, as well as for twelve months after the intervention ends. We will create a variable that captures the share of prescribed pills that were actually taken (i.e. number of actual bottle openings divided by the prescribed bottle openings).

Our biological primary outcome will be a binary indicator for whether the participant has an undetectable viral load. The AIDS Clinical Trials Group (ACTG) defines virologic failure as a confirmed viral load >200 copies/mL. We will indicate a viral load measure as undetectable if it is below 200 copies/mL. Viral load is the primary measure used to assess level of viral activity in a person’s blood, as well as response to ART. Although other factors besides adherence contribute to viral load, undetectable viral load is widely considered a strong indicator of good ART adherence. Furthermore, given the limitations of measures of behavioral adherence, viral load is considered by some to be the best indicator of adherence, and at the very least a valuable complement to behavioral adherence measures. Viral load measurements are now part of routine clinical care at Mildmay and will be chart abstracted. They are taken when a person tests HIV-positive, six and twelve months thereafter, and then every 12 months. We will line up recruitment so that all participants have a viral load measure taken at baseline, around month 12, and around month 24.
Secondary Outcomes
Secondary Outcomes (end points)
Viral Load as continuous variable, and measure of adherence (90%)
Secondary Outcomes (explanation)
First, as an additional measure of viral load, we will analyze viral load as a log-transformed continuous variable instead of a binary variable indicating detectable/undetectable. Second, as an additional measure of adherence, we will create a binary indicator for whether the client opened 90% or more of the openings that were prescribed. We will also create an indicator for treatment interruptions of more than 48 hours, which is an important predictor of virologic failure.
Experimental Design
Experimental Design
This study will use a three-armed randomized controlled trial (two intervention groups and one control group), with 1:1:1 randomization at the individual level. The intervention will last for 24 months. The study will be conducted at Mildmay Uganda, an NGO with headquarters in Uganda’s capitals Kampala. Mildmay Uganda specializes in the provision of comprehensive HIV and AIDS prevention, care, and treatment services.
Experimental Design Details
Not available
Randomization Method
Random treatment assignment will occur after participants are recruited, but before they complete a baseline survey using a 1:1:1 ratio. We will stratify random assignment on age (under-25 or over-25 years of age), gender, marital status (married/cohabiting or unmarried), low CD4 count (below 200 or above 200), and viral load (detectable or undetectable). Stratified randomization is achieved by generating a separate block for each combination of covariates identifying which clients fall into each block. We will then randomize treatment assignment within each block. We will use the randtreat package in Stata15 to carry out the randomization procedure.
Randomization Unit
Client-level (i.e. individual)
Was the treatment clustered?
No
Experiment Characteristics
Sample size: planned number of clusters
330 treatment-mature clients and 70 treatment-initiating clients
Sample size: planned number of observations
330 treatment-mature clients and 70 treatment-initiating clients
Sample size (or number of clusters) by treatment arms
110 clients per arm, plus 70 treatment initiators in the T2 arm eligible for prize drawings conditional on high MEMS-cap measured adherence
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
We have calculated the size of effects that our sample will be able to detect with 80% power (2-tailed test) with regard to outcomes at month 12 and 24, and 10% attrition every year (we observed 5% attrition over 20 months in our previous studies so this is a conservative estimate). For the primary outcome of viral suppression, we use a conservative estimate of 70% of clients in the control group showing suppression based on discussions with the Mildmay team. Our sample size of 110 participants in each of the three arms (total n=330) will be able to detect a 7 percentage point difference for joint comparison of T1 and T2 against the control at month 12, and about an 8.5 percentage point difference between the two intervention arms (a ‘subgroup analysis’). The corresponding differences at month 24 are 8 and 9 percentage points, respectively. These are considered small effect sizes (Cohen’s d of between .15 and .185) that we will be able to detect. For adherence, in our previous study we observed mean adherence rates of ~ 75% as measured by MEMS-caps; our sample size of 110 in each of the intervention arms and the control group will provide sufficient power to detect about a 6.5 percentage point difference in mean adherence between the two intervention arms (combined) and the control group. To test for differences in adherence between the two intervention arms, our study is powered to detect about a 7.5 percentage point effect as measured by MEMS- caps. Again, this means we will be able to detect small-sized effects. The corresponding differences we will be able to detect at month 24 are about 7 and 8 percentage points, respectively.
IRB
INSTITUTIONAL REVIEW BOARDS (IRBs)
IRB Name
Uganda National Council for Science and Technology
IRB Approval Date
2018-04-07
IRB Approval Number
HS 2394
IRB Name
Mildmay Uganda Research Ethics Committee
IRB Approval Date
2018-04-03
IRB Approval Number
0203-2018
IRB Name
Human Subjects Protection Committee at RAND Corporation
IRB Approval Date
2018-03-02
IRB Approval Number
2016-0956