Synthetic Controls for Implementation Science: Opportunities for HIV Program Evaluation Using Routinely Collected Data.

PURPOSE OF REVIEW: HIV service delivery programs are some of the largest funded public health programs in the world. Timely, efficient evaluation of these programs can be enhanced with methodologies designed to estimate the effects of policy. We propose using the synthetic control method (SCM) as an implementation science tool to evaluate these HIV programs. RECENT FINDINGS: SCM, introduced in econometrics, shows increasing utility across fields. Key benefits of this methodology over traditional design-based approaches for evaluation stem from directly approximating pre-intervention trends by weighting of candidate non-intervention units. We demonstrate SCM to evaluate the effectiveness of a public health intervention targeting HIV health facilities with high numbers of recent infections on trends in pre-exposure prophylaxis (PrEP) enrollment. This test case demonstrates SCM's feasibility for effectiveness evaluations of site-level HIV interventions. HIV programs collecting longitudinal, routine service delivery data for many facilities, with only some receiving a time-specified intervention, are well-suited for evaluation using SCM.

Brief Report: Self-Reported HIV-Positive Status but Subsequent HIV-Negative Test Results in Population-Based HIV Impact Assessment Survey Participants-11 Sub-Saharan African Countries, 2015-2018.

BACKGROUND: HIV testing is a critical step to accessing antiretroviral therapy (ART) because early diagnosis can facilitate earlier initiation of ART. This study presents aggregated data of individuals who self-reported being HIV-positive but subsequently tested HIV-negative during nationally representative Population-Based HIV Impact Assessment surveys conducted in 11 countries from 2015 to 2018. METHOD: Survey participants aged 15 years or older were interviewed by trained personnel using a standard questionnaire to determine HIV testing history and self-reported HIV status. Home-based HIV testing and counseling using rapid diagnostic tests with return of results were performed by survey staff according to the respective national HIV testing services algorithms on venous blood samples. Laboratory-based confirmatory HIV testing for all participants identified as HIV-positives and self-reported positives, irrespective of HIV testing results, was conducted and included Geenius HIV-1/2 and DNA polymerase chain reaction if Geenius was negative or indeterminate. RESULTS: Of the 16,630 participants who self-reported as HIV-positive, 16,432 (98.6%) were confirmed as HIV-positive and 198 (1.4%) were HIV-negative by subsequent laboratory-based testing. Participants who self-reported as HIV-positive but tested HIV-negative were significantly younger than 30 years, less likely to have received ART, and less likely to have received a CD4 test compared with participants who self-reported as HIV-positive with laboratory-confirmed infection. CONCLUSIONS: A small proportion of self-reported HIV-positive individuals could not be confirmed as positive, which could be due to initial misdiagnosis, deliberate wrong self-report, or misunderstanding of the questionnaire. As universal ART access is expanding, it is increasingly important to ensure quality of HIV testing and confirmation of HIV diagnosis before ART initiation.

Point of Care CD4 Testing in National Household Surveys - Results and Quality Indicators from Eleven Population-Based HIV Impact Assessment (PHIA) Surveys.

Population-based HIV Impact Assessments (PHIAs) are national household (HH) surveys that provide HIV diagnosis and CD4 testing with an immediate return of results. Accurate CD4 results improve HIV-positive participants' clinical care and inform the effectiveness of HIV programs. Here, we present CD4 results from the PHIA surveys that were conducted in 11 countries in sub-Saharan Africa between 2015 and 2018. All of the HIV-positive participants and 2 to 5% of the HIV-negative participants were offered Pima CD4 (Abbott, IL, USA) point-of-care (POC) tests. The quality of the CD4 test was ensured by conducting instrument verification, comprehensive training, quality control, a review of testing errors and an analysis of unweighted CD4 data by HIV status, age, gender, and antiretroviral (ARV) treatment status. Overall, CD4 testing was completed for 23,085 (99.5%) of the 23,209 HIV-positive and 7,329 (2.7%) of the 270,741 negative participants in 11 surveys. The instrument error rate was 11.3% (range, 4.4% to 15.7%). The median CD4 values among HIV-positive and HIV-negative participants (aged 15+) were 468 cells/mm3 (interquartile range [IQR], 307 to 654) and 811 cells/mm3 (IQR, 647 to 1,013), respectively. Among the HIV-positive participants (aged 15+), those with detectable ARVs had higher CD4 values (508 cells/mm3) than those with undetectable ARVs (385.5 cells/mm3). Among the HIV-positive participants (aged 15+), 11.4% (2,528/22,253) had a CD4 value of less than 200 cells/mm3, and approximately half of them (1,225/2,528 = 48.5%) had detectable ARVs, whereas 51.5% (1,303/2,528) had no detectable ARVs (P < 0.0001). We successfully implemented high quality POC CD4 testing using Pima instruments. Our data come from nationally representative surveys in 11 countries and provide unique insights regarding the CD4 distribution among HIV-positive individuals as well as the baseline CD4 values among HIV-negative individuals. IMPORTANCE The manuscript describes CD4 levels among HIV-positive individuals and baseline CD4 levels among HIV-negative individuals from 11 sub-Saharan countries, thereby highlighting the importance of CD4 markers in the context of the HIV epidemic. Despite increased ARV access in each country, advanced HIV disease (CD4 < 200 cells/mm3) persists among approximately 11% of HIV-positive individuals. Therefore, it is important that our findings are shared with the scientific community to assist with similar implementations of point-of-care testing and to conduct a review of HIV programmatic gaps.

Correcting for selection bias in HIV prevalence estimates: an application of sample selection models using data from population-based HIV surveys in seven sub-Saharan African countries.

INTRODUCTION: Population-based biomarker surveys are the gold standard for estimating HIV prevalence but are susceptible to substantial non-participation (up to 30%). Analytical missing data methods, including inverse-probability weighting (IPW) and multiple imputation (MI), are biased when data are missing-not-at-random, for example when people living with HIV more frequently decline participation. Heckman-type selection models can, under certain assumptions, recover unbiased prevalence estimates in such scenarios. METHODS: We pooled data from 142,706 participants aged 15-49 years from nationally representative cross-sectional Population-based HIV Impact Assessments in seven countries in sub-Saharan Africa, conducted between 2015 and 2018 in Tanzania, Uganda, Malawi, Zambia, Zimbabwe, Lesotho and Eswatini. We compared sex-stratified HIV prevalence estimates from unadjusted, IPW, MI and selection models, controlling for household and individual-level predictors of non-participation, and assessed the sensitivity of selection models to the copula function specifying the correlation between study participation and HIV status. RESULTS: In total, 84.1% of participants provided a blood sample to determine HIV serostatus (range: 76% in Malawi to 95% in Uganda). HIV prevalence estimates from selection models diverged from IPW and MI models by up to 5% in Lesotho, without substantial precision loss. In Tanzania, the IPW model yielded lower HIV prevalence estimates among males than the best-fitting copula selection model (3.8% vs. 7.9%). CONCLUSIONS: We demonstrate how HIV prevalence estimates from selection models can differ from those obtained under missing-at-random assumptions. Further benefits include exploration of plausible relationships between participation and outcome. While selection models require additional assumptions and careful specification, they are an important tool for triangulating prevalence estimates in surveys with substantial missing data due to non-participation.

Prevalence of and factors associated with late diagnosis of HIV in Malawi, Zambia, and Zimbabwe: Results from population-based nationally representative surveys.

INTRODUCTION: Late diagnosis of HIV (LD) increases the risk of morbidity, mortality, and HIV transmission. We used nationally representative data from population-based HIV impact assessment (PHIA) surveys in Malawi, Zambia, and Zimbabwe (2015-2016) to characterize adults at risk of LD and to examine associations between LD and presumed HIV transmission to cohabiting sexual partners. METHODS: We estimated the prevalence of LD, defined as CD4 count <350 cells/µL, among adults newly diagnosed with HIV during the surveys and odds ratios for associated factors. We linked newly diagnosed adults (index cases) to their household sexual partners and calculated adjusted odds ratios for associations between LD of the index case, viral load of the index case, and duration of HIV exposure in the relationship, and the HIV status of the household sexual partner. RESULTS: Of 1,804 adults who were newly diagnosed with HIV in the surveys, 49% (882) were diagnosed late. LD was associated with male sex, older age, and almost five times the odds of having an HIV-positive household sexual partner (adjusted odds ratio [aOR], 4.65 [95% confidence interval: 2.56-8.45]). Longer duration of HIV exposure in a relationship and higher viral load of the index case were both independently associated with higher odds of having HIV-positive household sexual partners. Individuals with HIV exposure of more than 5 years had more than three times (aOR 3.42 [95% CI: 1.63-7.18]) higher odds of being HIV positive than those with less than 2 years HIV exposure. The odds of being HIV positive were increased in individuals who were in a relationship with an index case with a viral load of 400-3499 copies/mL (aOR 4.06 [95% CI 0.45-36.46]), 3,500-9,999 copies/mL (aOR 11.32 [95% CI: 4.08-31.39]), 10,000-49,999 copies/mL (aOR 17.07 [95% CI: 9.18-31.72]), and ≥50,000 copies/mL (aOR 28.41 [95% CI: 12.18-66.28]) compared to individuals who were in a relationship with an index case with a viral load of <400 copies/mL. CONCLUSIONS: LD remains a challenge in Southern Africa and is strongly associated with presumed HIV transmission to household sexual partners. Our study underscores the need for earlier HIV diagnosis, particularly among men and older adults, and the importance of index testing.

HIV incidence, viremia, and the national response in Eswatini: Two sequential population-based surveys.

With the highest HIV incidence and prevalence globally, the government of Eswatini started a substantial scale-up of HIV treatment and prevention services in 2011. Two sequential large population-based surveys were conducted before and after service expansion to assess the impact of the national response. Cross-sectional, household-based, nationally representative samples of adults, ages 18 to 49 years, were sampled in 2011 and 2016. We measured HIV prevalence, incidence (recent infection based on limiting antigen ≤1.5 optical density units and HIV RNA ≥1000 copies/mL), viral load suppression (HIV RNA <1000 copies/mL among all seropositive adults) and unsuppressed viremia (HIV RNA ≥1000 copies/mL among all, regardless of HIV status) and assessed for temporal changes by conducting a trend analysis of the log ratio of proportions, using a Z statistic distribution. HIV prevalence remained stable from 2011 to 2016 [32% versus 30%, p = 0.10]. HIV incidence significantly declined 48% [2.48% versus 1.30%, p = 0.01]. Incidence remained higher among women than men [2011: 3.16% versus 1.83%; 2016: 1.76% versus 0.86%], with a smaller but significant relative reduction among women [44%; p = 0.04] than men [53%; p = 0.09]. The proportion of seropositive adults with viral load suppression significantly increased from 35% to 71% [p < .001]. The proportion of the total adult population with unsuppressed viremia decreased from 21% to 9% [p < .001]. National HIV incidence in Eswatini decreased by nearly half and viral load suppression doubled over a five-year period. Unsuppressed viremia in the total population decreased 58%. These population-based findings demonstrate the national impact of expanded HIV services in a hyperendemic country.

Opportunities for Closing the Gap in HIV Diagnosis, Treatment, and Viral Load Suppression in Children in Malawi: Results From a 2015-2016 Population-based HIV Impact Assessment Survey.

BACKGROUND: Control of the pediatric HIV epidemic is hampered by gaps in diagnosis and linkage to effective treatment. The 2015-2016 Malawi Population-based HIV impact assessment data were analyzed to identify gaps in pediatric HIV diagnosis, treatment, and viral load suppression. METHODS: In half of the surveyed households, children ages ≥18 months to <15 years were tested using the national HIV rapid test algorithm. Children ≤18 months reactive by the initial rapid test underwent HIV total nucleic acid polymerase chain reaction confirmatory testing. Blood from HIV-positive children was tested for viral load (VL) and presence of antiretroviral drugs. HIV diagnosis and antiretroviral treatment (ART) use were defined using guardian-reporting or antiretroviral detection. RESULTS: Of the 6166 children tested, 99 were HIV-positive for a prevalence of 1.5% (95% confidence intervals [CI]: 1.1-1.9) and 8.0% (95% CI: 5.6-10.5) among HIV-exposed children. The prevalence of 1.5% was extrapolated to a national estimate of 119,501 (95% CI: 89,028-149,974) children living with HIV (CLHIV), of whom, 30.7% (95% CI: 20.3-41.1) were previously undiagnosed. Of the 69.3% diagnosed CLHIV, 86.1% (95% CI: 76.8-95.6) were on ART and 57.9% (95% CI: 41.4-74.4) of those on ART had suppressed VL (<1000 HIV RNA copies/mL). Among all CLHIV, irrespective of HIV diagnosis or ART use, 57.7% (95% CI: 45.0-70.5) had unsuppressed VL. CONCLUSIONS: Critical gaps in HIV diagnosis in children persist in Malawi. The large proportion of CLHIV with unsuppressed VL reflects gaps in diagnosis and need for more effective first- and second-line ART regimens and adherence interventions.

Progress Toward the 90-90-90 HIV Targets in Zimbabwe and Identifying Those Left Behind.

OBJECTIVE: We present findings from the nationally representative Zimbabwe Population-based HIV Impact Assessment that characterize Zimbabwe's progress toward the Joint United Nations Programme on HIV/AIDS 90-90-90 targets. DESIGN: We conducted a cross-sectional household survey. METHODS: Consenting adults and children in the household were eligible to participate in Zimbabwe Population-based HIV Impact Assessment (October 2015-August 2016). Participants completed face-to-face interviews and provided blood for HIV, CD4, viral load, and syphilis testing. Viral load suppression (VLS) was defined as HIV RNA <1000 copies/mL. HIV-positive specimens were tested for the presence of selected antiretroviral drugs. Data were weighted. Analysis was restricted to HIV-positive adults aged 15-64 years. RESULTS: We enrolled 11,098 men and 14,033 women aged 15-64 years. HIV prevalence was 14.1%. Of those living with HIV, 76.8% (95% confidence interval [CI]: 74.9 to 78.7) were aware of their HIV status or had detectable antiretroviral levels. Of these, 88.4% (95% CI: 87.1 to 89.7) were receiving antiretroviral therapy (ART), and of these people, 85.3% (95% CI: 83.4 to 87.1) had VLS. Male sex age 15-34 years and having 1 or more sexual partners were associated with being unaware of one's HIV-positive status. Age <50 years and not taking cotrimoxazole were associated with being less likely to be being both aware and taking ART. Male sex, age <50 years, and taking cotrimoxazole were associated with being on ART but not having VLS. CONCLUSIONS: Zimbabwe has made great strides toward epidemic control. Focusing resources on case finding, particularly among men, people aged <35 years, and sexually active individuals can help Zimbabwe attain 90-90-90 targets.

Population-Based HIV Impact Assessments Survey Methods, Response, and Quality in Zimbabwe, Malawi, and Zambia.

BACKGROUND: The population-based HIV impact assessment (population-based HIV impact assessments) surveys are among the first to estimate national adult HIV incidence, subnational prevalence of viral load suppression, and pediatric HIV prevalence. We summarize the survey methods implemented in Zimbabwe, Malawi, and Zambia, as well as response rates and quality metrics. METHODS: Each cross-sectional, household-based survey used a 2-stage cluster design. Survey preparations included sample design, questionnaire development, tablet programming for informed consent and data collection, community mobilization, establishing a network of satellite laboratories, and fieldworker training. Interviewers collected demographic, behavioral, and clinical information using tablets. Blood was collected for home-based HIV testing and counseling (HBTC) and point-of-care CD4+ T-cell enumeration with results immediately returned. HIV-positive blood samples underwent laboratory-based confirmatory testing, HIV incidence testing, RNA polymerase chain reaction (viral load), DNA polymerase chain reaction (early infant diagnosis), and serum antiretroviral drug detection. Data were weighted for survey design, and chi square automatic interaction detection-based methods were used to adjust for nonresponse. RESULTS: Each survey recruited a nationally representative, household-based sample of children and adults over a 6-10-month period in 2015 and 2016. Most (84%-90%) of the 12,000-14,000 eligible households in each country participated in the survey, with 77%-81% of eligible adults completing an interview and providing blood for HIV testing. Among eligible children, 59%-73% completed HIV testing. Across the 3 surveys, 97.8% of interview data were complete and had no errors. CONCLUSION: Conducting a national population-based HIV impact assessment with immediate return of HIV and other point-of-care test results was feasible, and data quality was high.

A Comprehensive Approach to Assuring Quality of Laboratory Testing in HIV Surveys: Lessons Learned From the Population-Based HIV Impact Assessment Project.

BACKGROUND: Conducting HIV surveys in resource-limited settings is challenging because of logistics, limited availability of trained personnel, and complexity of testing. We described the procedures and systems deemed critical to ensure high-quality laboratory data in the population-based HIV impact assessments and large-scale household surveys. METHODS: Laboratory professionals were engaged in every stage of the surveys, including protocol development, site assessments, procurement, training, quality assurance, monitoring, analysis, and reporting writing. A tiered network of household, satellite laboratories, and central laboratories, accompanied with trainings, optimized process for blood specimen collection, storage, transport, and real-time monitoring of specimen quality, and test results at each level proved critical in maintaining specimen integrity and high-quality testing. A plausibility review of aggregate merged data was conducted to confirm associations between key variables as a final quality check for quality of laboratory results. RESULTS: Overall, we conducted a hands-on training for 3355 survey staff across 13 surveys, with 160-387 personnel trained per survey on biomarker processes. Extensive training and monitoring demonstrated that overall, 99% of specimens had adequate volume and 99.8% had no hemolysis, indicating high quality. We implemented quality control and proficiency testing for testing, resolved discrepancies, verified >300 Pima CD4 instruments, and monitored user errors. Aggregate data review for plausibility further confirmed the high quality of testing. CONCLUSIONS: Ongoing engagement of laboratory personnel to oversee processes at all levels of the surveys is critical for successful national surveys. High-quality population-based HIV impact assessments laboratory data ensured reliable results and demonstrated the impact of HIV programs in 13 countries.