HIV
Conditions
Keywords
HIV, Food insecurity, Microcredit loan, Agriculture, Livelihoods
Brief summary
The purpose of this study is to determine whether this multisectoral agricultural and microcredit loan intervention improves food security, prevent antiretroviral treatment failure, and reduce co-morbidities among people living with HIV/AIDS.
Detailed description
Despite major advances in care and treatment for those living with HIV, morbidity and mortality among people living with HIV/AIDS (PLHIV) remains unacceptably high in sub-Saharan Africa (SSA), largely due to the parallel challenges of poverty and food insecurity.\[1\] In the Nyanza Region of Kenya, 15.1% of the adult population is infected by HIV,\[2\] and over 50% of the rural population is food insecure, primarily due to unpredictable rainfall and limited irrigation.\[3,4\] The investigators have previously shown that food insecurity delays antiretroviral therapy (ART) initiation, reduces ART adherence, contributes to worse immunologic and virologic outcomes, and increases morbidity and mortality among PLHIV.\[5-16\] There has been increasing international recognition that improved food security and reduced poverty are essential components for a successful global response to the HIV epidemic.\[17-21\] Yet, to date few studies have systematically evaluated the impacts of sustainable food security interventions on health, economic, and behavioral outcomes among PLHIV. Agricultural interventions, which have potential to raise income and bolster food security, are an important but understudied route through which to sustainably improve nutritional and HIV outcomes in SSA, including Kenya where agriculture accounts for \> 75% of the total workforce, and 51% of the gross domestic product.\[22\] Building on the investigators successful completion of the pilot intervention trial in Kenya and the investigators collective experience studying structural barriers to HIV care in SSA, the investigators plan to test the hypothesis that a multisectoral agricultural and microcredit loan intervention will improve food security, prevent ART treatment failure, and reduce co-morbidities among PLHIV. The investigators' intervention was co-developed with KickStart, a prominent non-governmental organization (NGO) based in SSA that has introduced a human-powered pump, enabling farmers to grow high yield crops year-round. This technology has reduced food insecurity and poverty for 800,000 users in 22 countries in the subcontinent since 1991.\[23\] The investigators' intervention includes: a) a loan (\ $175) from a well-established Kenyan bank for purchasing agricultural implements and commodities; b) agricultural implements to be purchased with the loan including the KickStart treadle pump, seeds, fertilizers and pesticides; and c) education in financial management and sustainable farming practices occurring in the setting of patient support groups. This study is a cluster randomized controlled trial (RCT) of this intervention with the following specific aims: Aim 1: To determine the impact of a multisectoral agricultural intervention among HIV-infected farmers on ART on HIV clinical outcomes. The investigators hypothesize that the intervention will lead to improved viral load suppression (primary outcome) and changes in CD4 cell count, physical health status, WHO stage III/IV disease, and hospitalizations (secondary outcomes) in the intervention arm compared to the control arm. Aim 2: To understand the pathways through which the multisectoral intervention may improve HIV health outcomes. Using the investigator's theoretical model,\[1,24\] the investigators hypothesize that the intervention will improve food security and household wealth, which in turn will contribute to improved outcomes through nutritional (improved nutritional status measured with Body Mass Index), behavioral (improved ART adherence, and retention in care), and mental health (improved mental health/less depression, improved empowerment) pathways (secondary outcomes). Aim 3: To determine the cost-effectiveness of the intervention and obtain the information necessary to inform scale-up in Kenya and similar settings in SSA. The investigators will quantify the cost per disability-adjusted life year averted, and identify lessons to inform successful scale-up. To accomplish Aims 1 & 2, the investigators will randomize 8 matched pairs of health facilities in the Nyanza Region in a 1:1 ratio to the intervention and control arms, and enroll 44 participants per facility (total n=704). All participants will be followed for 2 years. Impacts of the investigator's intervention on primary health outcomes and mediators will be investigated to provide definitive data of direct and indirect intervention effects. To accomplish Aim 3, the investigators will: a) conduct a cost-effectiveness analysis; b) identify the characteristics of individuals most likely to benefit from the intervention (e.g., gender, educational attainment, family size, wealth, risk tolerance, and entrepreneurial ability); and c) perform a mixed-methods process evaluation with study participants, staff, and various stakeholders to determine what worked and did not work to guide future scale-up efforts of the intervention. The investigator's ultimate goal is to develop and test an intervention to reverse the cycle of food insecurity and HIV/AIDS morbidity and mortality in SSA.
Interventions
OTHERShamba Maisha Intervention
A) A loan (\ $175) B) Agricultural implements to be purchased with the loan C) Education in financial management and sustainable farming practices
Sponsors
National Institute of Mental Health (NIMH)
Kenya Medical Research Institute
University of South Carolina
University of Connecticut
University of Pennsylvania
University of California, San Francisco
Study design
Allocation
RANDOMIZED
Intervention model
PARALLEL
Primary purpose
OTHER
Masking
NONE
Eligibility
Sex/Gender
ALL
Age
18 Years to 60 Years
Healthy volunteers
No
Inclusion criteria
* HIV-infected adults * Currently receiving ART * Belong to a patient support group or demonstrate willingness to join a support group * Agree to save the down payment (\ $10) required for the microcredit loan * Have evidence of moderate to severe food insecurity based on the Household Food Insecurity Access Scale (HFIAS), and/or malnutrition (BMI\<18.5) based on FACES medical records during the year preceding recruitment * Have access to farming land and available surface water in the form of lakes, rivers, ponds and shallow wells
Exclusion criteria
* People who do not speak Dholuo, Swahili, or English * Inadequate cognitive and/or hearing capacity to complete planned study procedures, at the discretion of the research assistant
Design outcomes
Primary
| Measure | Time frame | Description |
|---|---|---|
| Change in Proportion of Viral Load Suppression (<=200 Copies/mL) | Baseline and endline (2 years after enrollment) | The outcome was the change from baseline to the end of follow-up (2 years) in the proportion of participants in viral load suppression (≤200 copies/mL) compared between study groups using difference-in-differences analyses. |
Secondary
| Measure | Time frame | Description |
|---|---|---|
| Change (i.e. Linear Trend) in Mean Physical Health Status | Baseline and endline (2 years after enrollment) | The outcome was the change (i.e. linear trend) from baseline to the end of follow-up (2 years) of the mean physical health score compared between study groups using the differences-in-differences analyses. We used the Medical Outcomes Study HIV Health Survey (MOS-HIV), a tool used to assess health-related quality of life that has been validated in resource-limited settings. Scores standardized to a range of 0 to 100. Higher scores mean a better outcome. |
| Change (i.e., Linear Trend) in the Proportion of Participants With AIDS-Defining Condition | Baseline and endline (2 years after enrollment) | The outcome was the change (i.e., linear trend) from baseline to the end of follow-up (2 years) of the proportion of participants with an AIDS-defining condition, compared between study groups using difference-in-differences analyses. AIDS-defining conditions including HIV-related illnesses included in the Centers for Disease Control and Prevention's (CDC) list of diagnostic criteria for AIDS. AIDS-defining conditions include opportunistic infections and cancers that are life-threatening in a person with HIV. |
| Change (i.e., Linear Trend) in the Proportion of Participants Who Were Hospitalized in the Previous 6 Months | Baseline and endline (2 years after enrollment) | The outcome was the change (i.e. linear trend) from baseline to end of follow-up (2 years) of the proportion of participants hospitalized in the previous 6 months (yes/no), compared between study groups using difference-in-differences analyses. |
| Change (i.e. Linear Trend) in the Mean Score of Food Insecurity Score | Baseline and endline (2 years after enrollment) | The outcome was the change (i.e. linear trend) from baseline to the end of follow-up (2 years) of the mean household food insecurity score, compared between study groups using difference-in-differences analyses. using the Household Food Insecurity Access Scale (HFIAS). The HFIAS is a tool to assess household food insecurity (access). The scale scores range from 0 to 27, with higher scores indicating greater food insecurity. |
| Change (i.e., Linear Trend) in Proportion of Absolute CD4 Count <=500 Cells/mm^3 | Baseline and endline (2 years after enrollment) | The outcome was the change (i.e., linear trend) from baseline to the end of follow-up (2 years) of the proportion of participants with a CD4 cell count \<=500 cells/mm\^3, compared between study groups using difference-in-differences analyses. |
| Change (i.e. Linear Trend) in Mean Self-reported Adherence to Antiretroviral Therapy | Baseline and endline (2 years after enrollment) | The outcome was the change (i.e. linear trend) from baseline to the end of follow-up (2 years) of the mean self-reported adherence to antiretroviral therapy compared between study groups using the differences-in-differences analyses. |
| Change (i.e. Linear Trend) in Mean Self-confidence Score | Baseline and endline (2 years after enrollment) | The outcome was the change (i.e. linear tend) from baseline to the end of follow-up (2 years) in the mean self-confidence score, compared between study groups using difference-in-differences analyses. Self-confidence is measured using the three-item Power Within scale, which has a range of 3 to 9 points where lower scores indicate greater self-confidence. |
| Change (i.e. Linear Trend) in Proportion of Probable Depression | Baseline and endline (2 years after enrollment) | The outcome was the change (i.e. linear trend) from baseline to the end of follow-up (2 years) in the proportion with probable depression using the Hopkins Symptom Check-list for Depression, compared between study groups using difference-in-differences analyses. |
| Change (i.e. Linear Trend) in the Mean Internalized Stigma Score | Baseline and endline (2 years after enrollment) | The outcome was the change (i.e. linear trend) from baseline to the end of follow-up (2 years) in the mean internalized stigma score compared between study groups using the differences-in-differences analyses. Internalized HIV stigma arises when someone has accepted and endorsed the negative attitudes towards her/himself due to their HIV status. The internalized HIV stigma sub-scale consisted of six items asking respondents to agree with statements related to how they feel about being HIV positive, such as having HIV makes me feel like I'm a bad person and I feel ashamed of having HIV. Response options ranged from 1 strongly disagree to 5 strongly agree. During the analysis phase, the composite scores of each stigma sub-scale were rescaled to a row average of 1-5, with higher scores indicating greater stigma. |
| Change (i.e. Linear Trend) in Mean Nutritional Status (Represented by Body Mass Index (BMI)) | Baseline and endline (2 years after enrollment) | The outcome was the change (i.e. linear trend) from baseline to the end of follow-up (2 years) of the mean body mass index (BMI) compared between study grouops using the differences-in-differences analyses. |
Countries
Kenya
Participant flow
Participants by arm
| Arm | Count |
|---|---|
| Intervention Participants received the multisectoral agricultural intervention. | 366 |
| Control Participants in the control arm received the standard of care. | 354 |
| Total | 720 |
Withdrawals & dropouts
| Period | Reason | FG000 | FG001 |
|---|---|---|---|
| Overall Study | Death | 5 | 4 |
| Overall Study | Did not meet enrollment criteria | 0 | 1 |
| Overall Study | Hospitalized before receipt of any study activities | 1 | 0 |
| Overall Study | Imprisoned | 0 | 2 |
| Overall Study | Lost to Follow-up | 7 | 4 |
| Overall Study | Moved out of study area before receiving any study activities | 9 | 11 |
| Overall Study | Non-payment of loan down-payment | 16 | 0 |
| Overall Study | Uncomfortable with MEMS adherence monitoring cap | 1 | 0 |
| Overall Study | Withdrawal by Subject | 7 | 1 |
Baseline characteristics
| Characteristic | Control | Intervention | Total |
|---|---|---|---|
| Age, Continuous | 40.4 years STANDARD_DEVIATION 9.3 | 40.3 years STANDARD_DEVIATION 8.9 | 40.4 years STANDARD_DEVIATION 9.1 |
| BMI <18.5 kg/m^2 | 45 Participants | 41 Participants | 86 Participants |
| CD4+ | 561 cells/mm^3 STANDARD_DEVIATION 235 | 603 cells/mm^3 STANDARD_DEVIATION 276 | 582 cells/mm^3 STANDARD_DEVIATION 257 |
| Currently married | 251 Participants | 271 Participants | 522 Participants |
| Number of people in household | 6.1 people STANDARD_DEVIATION 2.7 | 6.5 people STANDARD_DEVIATION 2.6 | 6.3 people STANDARD_DEVIATION 2.7 |
| Race/Ethnicity, Customized Black Kenyan | 354 Participants | 366 Participants | 720 Participants |
| Region of Enrollment Kenya | 354 Participants | 366 Participants | 720 Participants |
| Religion Catholic | 353 Participants | 363 Participants | 716 Participants |
| Religion Muslim | 1 Participants | 0 Participants | 1 Participants |
| Religion Other religion | 0 Participants | 3 Participants | 3 Participants |
| Severely food insecure (vs. moderately) | 275 Participants | 293 Participants | 568 Participants |
| Sex: Female, Male Female | 194 Participants | 202 Participants | 396 Participants |
| Sex: Female, Male Male | 160 Participants | 164 Participants | 324 Participants |
| Viral load <=200 copies/mL | 291 Participants | 314 Participants | 605 Participants |
Adverse events
| Event type | EG000 affected / at risk | EG001 affected / at risk |
|---|---|---|
| deaths Total, all-cause mortality | 5 / 366 | 4 / 354 |
| other Total, other adverse events | 0 / 0 | 0 / 0 |
| serious Total, serious adverse events | 0 / 0 | 0 / 0 |
Outcome results
Primary
Change in Proportion of Viral Load Suppression (<=200 Copies/mL)
The outcome was the change from baseline to the end of follow-up (2 years) in the proportion of participants in viral load suppression (≤200 copies/mL) compared between study groups using difference-in-differences analyses.
Time frame: Baseline and endline (2 years after enrollment)
Population: 677 participants who completed endline data collection
| Arm | Measure | Value (COUNT_OF_PARTICIPANTS) |
|---|---|---|
| Intervention | Change in Proportion of Viral Load Suppression (<=200 Copies/mL) | 327 Participants |
| Control | Change in Proportion of Viral Load Suppression (<=200 Copies/mL) | 314 Participants |
Comparison: The investigators produced a difference-in-difference estimate between baseline and year 2 using multi-level logistic regression. Units are in log odds.p-value: 0.8695% CI: [-1, 0.84]Mixed Models Analysis
Secondary
Change (i.e. Linear Trend) in Mean Nutritional Status (Represented by Body Mass Index (BMI))
The outcome was the change (i.e. linear trend) from baseline to the end of follow-up (2 years) of the mean body mass index (BMI) compared between study grouops using the differences-in-differences analyses.
Time frame: Baseline and endline (2 years after enrollment)
Population: 677 participants who completed endline data collection
| Arm | Measure | Value (MEAN) |
|---|---|---|
| Intervention | Change (i.e. Linear Trend) in Mean Nutritional Status (Represented by Body Mass Index (BMI)) | 22.1 kg/m^2 |
| Control | Change (i.e. Linear Trend) in Mean Nutritional Status (Represented by Body Mass Index (BMI)) | 21.8 kg/m^2 |
Comparison: The investigators analyzed longitudinal data measured every 6 months to assess trends in the two study arms and produced a difference-in-differences estimate between baseline and year 2.p-value: 0.0295% CI: [-0.39, -0.04]Mixed Models Analysis
Secondary
Change (i.e. Linear Trend) in Mean Physical Health Status
The outcome was the change (i.e. linear trend) from baseline to the end of follow-up (2 years) of the mean physical health score compared between study groups using the differences-in-differences analyses. We used the Medical Outcomes Study HIV Health Survey (MOS-HIV), a tool used to assess health-related quality of life that has been validated in resource-limited settings. Scores standardized to a range of 0 to 100. Higher scores mean a better outcome.
Time frame: Baseline and endline (2 years after enrollment)
Population: 677 participants who completed endline data collection
| Arm | Measure | Value (MEAN) |
|---|---|---|
| Intervention | Change (i.e. Linear Trend) in Mean Physical Health Status | 86.0 units on a scale |
| Control | Change (i.e. Linear Trend) in Mean Physical Health Status | 86.2 units on a scale |
Comparison: The investigators analyzed longitudinal data measured every 6 months to assess trends in the two study arms and produced a difference-in-differences estimate between baseline and year 2.p-value: 0.2695% CI: [-1.01, 3.78]Mixed Models Analysis
Secondary
Change (i.e. Linear Trend) in Mean Self-confidence Score
The outcome was the change (i.e. linear tend) from baseline to the end of follow-up (2 years) in the mean self-confidence score, compared between study groups using difference-in-differences analyses. Self-confidence is measured using the three-item Power Within scale, which has a range of 3 to 9 points where lower scores indicate greater self-confidence.
Time frame: Baseline and endline (2 years after enrollment)
Population: 677 participants who completed endline data collection
| Arm | Measure | Value (MEAN) |
|---|---|---|
| Intervention | Change (i.e. Linear Trend) in Mean Self-confidence Score | 4.0 score on a scale |
| Control | Change (i.e. Linear Trend) in Mean Self-confidence Score | 4.0 score on a scale |
Comparison: The investigators analyzed longitudinal data measured every 6 months to assess trends in the two study arms and produced a difference-in-differences estimate between baseline and year 2.p-value: 0.00195% CI: [-0.59, -0.15]Mixed Models Analysis
Secondary
Change (i.e. Linear Trend) in Mean Self-reported Adherence to Antiretroviral Therapy
The outcome was the change (i.e. linear trend) from baseline to the end of follow-up (2 years) of the mean self-reported adherence to antiretroviral therapy compared between study groups using the differences-in-differences analyses.
Time frame: Baseline and endline (2 years after enrollment)
Population: 677 participants who completed endline data collection
| Arm | Measure | Value (MEAN) |
|---|---|---|
| Intervention | Change (i.e. Linear Trend) in Mean Self-reported Adherence to Antiretroviral Therapy | 100 percentage of doses taken |
| Control | Change (i.e. Linear Trend) in Mean Self-reported Adherence to Antiretroviral Therapy | 100 percentage of doses taken |
Comparison: The investigators analyzed longitudinal data measured every 6 months to assess trends in the two study arms and produced a difference-in-differences estimate between baseline and year 2.p-value: 0.9895% CI: [-0.82, 0.84]Mixed Models Analysis
Secondary
Change (i.e., Linear Trend) in Proportion of Absolute CD4 Count <=500 Cells/mm^3
The outcome was the change (i.e., linear trend) from baseline to the end of follow-up (2 years) of the proportion of participants with a CD4 cell count \<=500 cells/mm\^3, compared between study groups using difference-in-differences analyses.
Time frame: Baseline and endline (2 years after enrollment)
Population: 677 participants who completed endline data collection
| Arm | Measure | Value (COUNT_OF_PARTICIPANTS) |
|---|---|---|
| Intervention | Change (i.e., Linear Trend) in Proportion of Absolute CD4 Count <=500 Cells/mm^3 | 128 Participants |
| Control | Change (i.e., Linear Trend) in Proportion of Absolute CD4 Count <=500 Cells/mm^3 | 129 Participants |
Comparison: The investigators analyzed longitudinal data measured every 6 months to assess trends in the two study arms and produced a difference-in-differences estimate between baseline and year 2 using multi-level logistic regression. Units are in log odds.p-value: 0.2395% CI: [-0.28, 0.92]Mixed Models Analysis
Secondary
Change (i.e. Linear Trend) in Proportion of Probable Depression
The outcome was the change (i.e. linear trend) from baseline to the end of follow-up (2 years) in the proportion with probable depression using the Hopkins Symptom Check-list for Depression, compared between study groups using difference-in-differences analyses.
Time frame: Baseline and endline (2 years after enrollment)
Population: 677 participants who completed endline data collection
| Arm | Measure | Value (COUNT_OF_PARTICIPANTS) |
|---|---|---|
| Intervention | Change (i.e. Linear Trend) in Proportion of Probable Depression | 36 Participants |
| Control | Change (i.e. Linear Trend) in Proportion of Probable Depression | 41 Participants |
Comparison: The investigators analyzed longitudinal data measured every 6 months to assess trends in the two study arms and produced a difference-in-differences estimate between baseline and year 2 using multi-level logistic regression. Units are in log odds.p-value: 0.00195% CI: [-1.45, -0.2]Mixed Models Analysis
Secondary
Change (i.e. Linear Trend) in the Mean Internalized Stigma Score
The outcome was the change (i.e. linear trend) from baseline to the end of follow-up (2 years) in the mean internalized stigma score compared between study groups using the differences-in-differences analyses. Internalized HIV stigma arises when someone has accepted and endorsed the negative attitudes towards her/himself due to their HIV status. The internalized HIV stigma sub-scale consisted of six items asking respondents to agree with statements related to how they feel about being HIV positive, such as having HIV makes me feel like I'm a bad person and I feel ashamed of having HIV. Response options ranged from 1 strongly disagree to 5 strongly agree. During the analysis phase, the composite scores of each stigma sub-scale were rescaled to a row average of 1-5, with higher scores indicating greater stigma.
Time frame: Baseline and endline (2 years after enrollment)
| Arm | Measure | Value (MEAN) | Dispersion |
|---|---|---|---|
| Intervention | Change (i.e. Linear Trend) in the Mean Internalized Stigma Score | 1.40 units on a scale | Standard Deviation 0.68 |
| Control | Change (i.e. Linear Trend) in the Mean Internalized Stigma Score | 1.72 units on a scale | Standard Deviation 0.71 |
p-value: <0.00195% CI: [-0.52, -0.31]Mixed Models Analysis
Secondary
Change (i.e. Linear Trend) in the Mean Score of Food Insecurity Score
The outcome was the change (i.e. linear trend) from baseline to the end of follow-up (2 years) of the mean household food insecurity score, compared between study groups using difference-in-differences analyses. using the Household Food Insecurity Access Scale (HFIAS). The HFIAS is a tool to assess household food insecurity (access). The scale scores range from 0 to 27, with higher scores indicating greater food insecurity.
Time frame: Baseline and endline (2 years after enrollment)
Population: 677 participants who completed endline data collection
| Arm | Measure | Value (MEAN) |
|---|---|---|
| Intervention | Change (i.e. Linear Trend) in the Mean Score of Food Insecurity Score | 14.0 score on a scale |
| Control | Change (i.e. Linear Trend) in the Mean Score of Food Insecurity Score | 15.0 score on a scale |
Comparison: The investigators analyzed longitudinal data measured every 6 months to assess trends in the two study arms and produced a difference-in-differences estimate between baseline and year 2.p-value: <0.00195% CI: [-4.16, -2.92]Mixed Models Analysis
Secondary
Change (i.e., Linear Trend) in the Proportion of Participants Who Were Hospitalized in the Previous 6 Months
The outcome was the change (i.e. linear trend) from baseline to end of follow-up (2 years) of the proportion of participants hospitalized in the previous 6 months (yes/no), compared between study groups using difference-in-differences analyses.
Time frame: Baseline and endline (2 years after enrollment)
Population: 677 participants who completed endline data collection
| Arm | Measure | Value (COUNT_OF_PARTICIPANTS) |
|---|---|---|
| Intervention | Change (i.e., Linear Trend) in the Proportion of Participants Who Were Hospitalized in the Previous 6 Months | 21 Participants |
| Control | Change (i.e., Linear Trend) in the Proportion of Participants Who Were Hospitalized in the Previous 6 Months | 18 Participants |
Comparison: The investigators analyzed longitudinal data measured every 6 months to assess trends in the two study arms and produced a difference-in-differences estimate between baseline and year 2 using multi-level logistic regression. Units are in log odds.p-value: 0.4495% CI: [-1.22, 0.53]Mixed Models Analysis
Secondary
Change (i.e., Linear Trend) in the Proportion of Participants With AIDS-Defining Condition
The outcome was the change (i.e., linear trend) from baseline to the end of follow-up (2 years) of the proportion of participants with an AIDS-defining condition, compared between study groups using difference-in-differences analyses. AIDS-defining conditions including HIV-related illnesses included in the Centers for Disease Control and Prevention's (CDC) list of diagnostic criteria for AIDS. AIDS-defining conditions include opportunistic infections and cancers that are life-threatening in a person with HIV.
Time frame: Baseline and endline (2 years after enrollment)
Population: 677 participants who completed endline data collection
| Arm | Measure | Value (COUNT_OF_PARTICIPANTS) |
|---|---|---|
| Intervention | Change (i.e., Linear Trend) in the Proportion of Participants With AIDS-Defining Condition | 4 Participants |
| Control | Change (i.e., Linear Trend) in the Proportion of Participants With AIDS-Defining Condition | 2 Participants |
Comparison: The investigators analyzed longitudinal data measured every 6 months to assess trends in the two study arms and produced a difference-in-differences estimate between baseline and year 2 using multi-level logistic regression. Units are in log odds.p-value: 0.3195% CI: [-0.79, 2.47]Mixed Models Analysis