What is PHIA?
The PHIA Project, led by ICAP at Columbia University in partnership with the US Centers for Disease Control and Prevention (CDC), measures the reach and impact of HIV programs in PEPFAR-supported countries through national surveys. Each population-based HIV impact assessment (PHIA) survey offers participants household-based HIV counseling and testing conducted by trained survey staff, with return of results. The surveys also asks questions about access to preventive care and treatment services for adults and children. The results will measure national and regional progress toward UNAIDS’ 90-90-90 goals and guide HIV policy and funding priorities.
ICAP also provides technical assistance at the national level to strengthen data collection systems, enhance laboratory infrastructure, and build capacity to design, conduct, and analyze PHIAs in survey countries.
The PHIA Project team includes: the University of California, San Francisco (UCSF), Global Health Sciences (GHS); ICF International; Westat; and the Statistical Center for HIV/AIDS Research and Prevention (SCHARP). Surveys are conducted in close partnership with ministries of health and local partners in each country.
The PHIA Project builds on ICAP’s expertise in data collection and management and lab-based measurement of HIV, as well as previous experience conducting SHIMS, a large-scale population survey of 13,000 households in Swaziland.
Frequently Asked Questions
The Population-based HIV Impact Assessment (PHIA) surveys are large, nationally representative household-based surveys that provide home-based HIV counseling and testing with immediate return of results. The surveys also:
- Collect data on access to and uptake of treatment and prevention services for HIV and the number of recent HIV infections (i.e., incidence)
- Test for CD4 T-cell counts and viral load suppression among HIV-positive participants
Data on treatment and prevention services is collected through household interviews, which include questions on behavioral factors associated with HIV. All information is recorded on tablets by trained survey staff and is treated as confidential.
The goal of the PHIA surveys is to provide a population-level understanding of the burden of HIV disease at the national and sub-national level in participating countries, and to document the achievements of HIV programs in these countries. PHIA surveys effectively serve as the HIV outcome and impact assessments stakeholders need to:
- Measure progress toward the UNAIDS 90-90-90 targets
- Measure HIV incidence and prevalence of viral load suppression, which will help stakeholders assess the effectiveness of past and current HIV prevention and treatment programs
By collecting data on the number of new HIV infections, the number of people living with HIV, and viral load suppression among people on treatment, the PHIA surveys will help governments understand how effective their HIV programs are and where future efforts should be focused.
PHIA surveys were first conducted in Malawi (MPHIA), Zambia (ZAMPHIA), and Zimbabwe (ZIMPHIA), with field implementation from late November 2015 to August 2016. Each PHIA is led by the national government through the ministry of health, conducted with funding from the U.S. President’s Emergency Plan for AIDS Relief (PEPFAR) and technical assistance from the U.S. Centers for Disease Control and Prevention (CDC). The surveys are being implemented by ICAP at Columbia University, in collaboration with local partners. For more information on where PHIA surveys are taking place, visit the Countries page.
Results of the PHIA Project will provide a valuable data resource for HIV and global health researchers, policymakers, and program planners, among others. The data can be used to inform evidence-based national and global HIV policies and funding. The information can also help guide program planning to focus on the most effective interventions and identify the areas and populations most in need of services.
For more information, visit the Data page.
PHIA surveys use a home-based testing and counseling (HBTC) approach to measure HIV prevalence. Specifically:
- Whole blood samples from venipuncture (adults and children over two years) or finger/heel stick (children under two years) are collected in the home.
- Blood samples are tested in the household using the national rapid HIV test algorithm. All HIV-positive samples are retested with a confirmatory test.
- For quality control, five percent of the HIV-negative samples are retested using the same national algorithm and confirmatory testing in a satellite laboratory.
For more information, visit the Methodology page.
Participants are asked to consent to an individual questionnaire, followed by a request to consent to HIV testing, including return of results through HBTC and other biomarker tests. Only participants who consent to testing and receiving results are included in the biomarker collection.
The DHS and PHIA surveys are both population-based surveys. However, there are key differences between these two types of surveys:
- The main purpose of the DHS is to provide data needed to monitor and evaluate population, health, and nutrition programs. In general, the DHS covers household and respondent characteristics, and questions that address:
- Fertility and family planning
- Infant and child health and mortality
- Maternal health and maternal and adult mortality
- Child and adult nutrition
- Domestic violence
- Orphans and vulnerable children.
Many DHS surveys do measure HIV prevalence and , but they generally do not return results. DHS, dissimilar to PHIA surveys, generally do not include children, may limit the adult age band differently, and may include only a sub-sample of adult males.
- The main purpose of the PHIA surveys is to estimate HIV incidence and prevalence. PHIA surveys collect data on the uptake of care and treatment services for HIV and other infectious diseases, measure CD4 T-cell counts, provide real-time HBTC, and estimate HIV incidence, viral load suppression, and HIV prevalence.
- Another important difference between the two surveys is the HIV testing method:
- The DHS approach generally uses anonymous dried blood spot (DBS) testing, with samples collected via finger prick.
- PHIA collects whole blood samples from participants in the home (through venipuncture for adults and children over two years and finger/heel stick for children under two years of age). PHIA uses the national rapid HIV test algorithm in the field and then performs confirmatory testing of HIV-positive samples with a confirmatory test (usually Geenius).
- In the past, DHS relied on ELISA testing without a confirmatory test. ELISA tests are highly sensitive and may identify as HIV-positive some people who are truly HIV-negative. If not followed by a confirmatory test, the ELISA may result in an overestimate of HIV prevalence. Therefore, pre-2015 DHS estimates may be falsely elevated compared to true prevalence.
When data are available, participation rates and HIV prevalence results will be directly compared between PHIA and DHS. Further analysis is needed to determine if the PHIA approach of HBTC results in a systematic participation bias.
PHIA HIV prevalence and incidence results from Malawi, Zambia, and Zimbabwe (released in December 2016) are similar to 2015 UNAIDS HIV estimates. Spectrum modeling estimates will be updated in 2017 and will use prevalence estimates from multiple sources, including results from both PHIA and DHS surveys.
In all PHIA countries to date, it has been challenging to find men at the household, even with three repeat visits. However, when men were present at the household, they were as likely as women to participate in the biomarker collection and HIV testing. The PHIA survey teams employed various strategies to try to reach men, such as:
- Offering the survey at a location of their choice (not necessarily at the household)
- Tailoring messages to mobilize men to participate in PHIA
- Conducting additional visits, after the completion of data collection, to households where men were not available for survey enrollment during the first three visits
PHIA survey data are weighted to account for non-response to the household interview, individual interview, and HIV testing. For household interview non-response, the weights of responding households are simply increased to account for any non-responding households within each enumeration area. For household interview and HIV testing non-response, a CHAID (CHi-squared Automatic Interaction Detector) algorithm is employed to adjust the weights of responding individuals to account for any non-responding individuals, based on characteristics of participants and non-participants. Finally, PHIA survey data are also weighted to account for non-coverage, which may arise due to incomplete enumeration of households or household members. This procedure involves calibrating non-response-adjusted weights so that the sum of the weights conforms to population totals from current population projection data.
The PHIA sample size in each country is designed to estimate national HIV incidence with a reasonable degree of precision among adults. As the number of recent infections is relatively low, further disaggregation by age or zone results in imprecise estimates with very wide confidence intervals. It is therefore not possible to make meaningful comparisons across these estimates.
The incidence algorithm used is recommended by the WHO Technical Working Group on HIV Incidence Assays. PHIA surveys use the HIV-1 Limiting Antigen (LAg) Avidity EIA to test all HIV-positive samples for recent infection. Results from the LAg are combined in an algorithm with viral load results to classify samples as recent or long-term infections (see Figure 1). Individuals who have recent infections according to LAg, but are also virally suppressed, are either elite controllers or on antiretroviral therapy (ART), which may result in a false recent infection result. Combining the LAg assay with viral load testing ensures that the LAg false-recents are properly classified as having long-term infection. HIV incidence estimates are calculated using the CDC Incidence Calculator, which uses the WHO-recommended equation for incidence calculations.
PHIA Incidence Testing Algorithm
The UNAIDS targets for HIV epidemic control are that 90 percent of all people living with HIV know their status, 90 percent of all people with diagnosed HIV infection received sustained ART, and 90 percent of all people receiving ART have viral suppression.
The PHIA surveys ask participants to self-report their HIV status (know they are HIV-negative, know they are HIV-positive, or don’t know their status) and, if positive, whether they are receiving ART. For those who consent to biomarker collection and test HIV-positive, a viral load test is conducted and individuals are classified as virally suppressed (<1000 copies/mL) or not suppressed. The 90-90-90 targets are then estimated as follows:
- First 90: Of those PHIA participants who tested HIV-positive, what percentage reported prior knowledge of their HIV-positive status?
- Second 90: Of those who reported their HIV-positive status, what percentage reported being on ART?
- Third 90: Of those who reported being on ART, what percentage were virally suppressed?
The “first report,” which includes the primary outcomes and some of the secondary outcome measures, is generally released five months after the end of data collection in a specific PHIA country. The final report and data set are generally released one year after the end of data collection in a specific PHIA country.
Participants who consent for biomarker collection and HIV testing undergo HBTC by trained counselors. The result of the HIV test rapid test algorithm, which is based on the national algorithm, is provided to the participants at the household during post-test counseling. In addition, point-of-care CD4 testing is performed and the results are provided to the participants at the household. HIV-positive participants are counseled to seek care at a care and treatment facility using an HIV referral form provided by the PHIA teams, particularly if they report that they are not already in HIV care or not already on ART.
Blood samples from HIV-positive participants are tested for viral load at a central laboratory. Children under 18 months with a positive rapid test result undergo Early Infant Diagnosis (EID) confirmatory testing. Both viral load and EID results are returned to the care and treatment facility chosen by the participant. Participants receive text messages when their viral load result and/or EID result of their child is available at the clinic. In some cases, PHIA teams go directly to the households to deliver EID results (whether positive or negative).
Additional centrally performed laboratory tests include HIV incidence testing (LAg Avidity), drug resistance testing, and testing to detect the presence of antiretrovirals in the blood. All clinically actionable results are returned to clinics.
The number of results communicated to participants in each facility is still being assessed.
Confidence Intervals: The 95 percent confidence intervals imply that if the PHIA survey was repeated many times in the country, on average 95 of every 100 surveys would contain the true population value. Even if the HIV prevalence point estimates differ between surveys, the estimates should be viewed in light of their confidence intervals. If intervals overlap, then any difference between the point estimates is not necessarily meaningful and could be due simply to chance.
Sample size: Generally, the sample size is designed to provide a national-level estimate of the annual HIV incidence rate among people aged 15 to 64 with a relative standard error of 30 percent or less, and also to provide regional-level estimates of viral load suppression among people aged 15 to 64 with a 95 percent confidence interval of +/- 10 percent or lower. These criteria are used to obtain an estimate of the number of households that need to be approached and take into account adjustments for household vacancy and non-response, number of individuals per household, individual non-response, and refusal of blood testing or specimen loss.
The PHIA surveys provide the first data on how far Malawi, Zambia, and Zimbabwe have come in terms of reaching epidemic control targets. They also support statistical modeling that shows declines in HIV incidence since 2003, which has previously only been estimated using modeled data. The surveys indicate that HIV prevalence is stabilizing, which suggests HIV-positive people are living longer due to accessible and effective treatment: approximately 68 percent of adults living with HIV in Malawi have viral suppression, compared to 60 percent in Zimbabwe, and 59 percent in Zambia. These results indicate that investments from national governments and other partners have had a significant impact, averting many thousands of new HIV infections and many thousands of AIDS-related deaths. The results also demonstrate where more attention is needed to meet the needs of particular populations, such as young women, girls, and adult men. These data can help governments and their partners improve and expand services to best serve the needs of their populations.
The PHIA surveys in Malawi, Zambia, and Zimbabwe (as well as other PHIA surveys) have collected data on behavior and other HIV risk factors, but these data have not yet been analyzed. The PHIA team is currently in the process of analyzing these data and findings will be presented and discussed in the first and final reports. Additionally, data sets will be made available in the Data section of this site approximately 12 months after the end of data collection in each specific PHIA country.