What we’re reading: Geospatial analysis of HIV

By on .

The following is a guest post by John Spencer of MEASURE Evaluation.

“How do we reach people needing HIV treatment?”

“What towns and communities should be prioritized to help stem the spread of HIV?”

Each year, PEPFAR teams around the world ask questions like these as they decide which programs to prioritize. It’s worth taking a moment to pause and think about these questions. Their simplicity belies the work, good-quality data, and strong decision tools needed to answer them.

The good news is that there is more data than ever before and the tools available to PEPFAR teams are getting stronger. Geospatial tools and methods in particular have seen exponential development—so much that it can be hard to keep up with the state of the art.

As part of MEASURE Evaluation’s global leadership in geosciences, we monitor developments in the field to identify innovations that might be applicable to PEPFAR and — more broadly — to global health. Recently we reviewed literature from the past two years related to geospatial targeting to identify emerging techniques and trends.

John Spencer

One noticeable trend is the wide range of methods and techniques that take advantage of the growing availability of health program and contextual data (demographic data, economic data, etc.). The type of analysis that can be done is considerable, and runs the gamut from basic traditional geographic techniques such as spatial aggregation and creation of choropleth maps, to more advanced approaches such as Bayesian modeling.

To help spread innovative techniques we’ve observed in the literature, we’re highlighting four articles representative of four approaches for geospatial analysis, presented in descending order of method complexity.

For more information and a collection of tools and resources for geospatial analysis methods and techniques visit MEASURE Evaluation.

Articles:

Geographical Inequalities and Social and Environmental Risk Factors for Under-Five Mortality in Ghana 2000 and 2010: Bayesian Spatial Analysis of Census Data – The authors used random samples from the 2000 and 2010 Ghanaian census, indirect demographic and Bayesian analysis to estimate under-five mortality at a fine geographic scale, and then assess the relationship between mortality and social and environmental risk factors. Covariates used are: cooking fuel, sanitation, drinking water source, parental education, rurality.  A key advantage of the approach is the ability to create subnational estimates using Bayesian methods linked with traditional multivariate analysis to gain a better understanding of risk factors. This has potential relevance with PEPFAR programming because subnational models of HIV incidence could be created to inform decision making relative to key risk factors. Others have used similar Bayesian approaches for HIV and the approach is reasonably well understood and implemented. The complexity of the approach, however, can be a disadvantage, as it requires specialized software and familiarity with Bayesian modeling. The resulting data products can sometimes be hard to interpret and require detailed explanation., Arku, Bennett, Castro,Agyeman-Duah, Mintah, Ware, Nyarko, Spengler,Agyei-Mensah, Ezzati, PLOS Medicine 13(6) 2016 .

Using geospatial mapping to design HIV elimination strategies for sub-Saharan Africa – A high-resolution density of infection map was produced for Lesotho and used to allocate a treatment approach that optimized efficiency by minimizing the target geographic area to find and treat HIV-infected individuals. The results were then used to evaluate the feasibility and potential for success for the 90-90-90 strategy. The advantage with this approach is that it is relatively easy to implement to look at the effectiveness of 90-90-90 strategy in a country. It enables modeling different treatment approaches to examine effects. For PEPFAR, a modified version has potential for understanding the implications of decisions around prioritizing low-volume sites. The disadvantage is the time required. Obtaining the data, cleaning it, and using it requires time, as does setting up implementation processes. This means it must be done in well in advance of any decision making deadline. An additional disadvantage is that it requires recent and accurate demographic and HIV data. Coburn, Okano, Blower,Science Translational Medicine 9 (2017) .

Assessing the spatial nonstationarity in relationship between local patterns of HIV infections and the covariates in South Africa: A geographically weighted regression analysis – Use of basic geographic weighted regression to explore the link between social covariates and HIV infections. An advantage is its simplicity. Geographic weighted regression is well known and easily executed, effective for understanding relationships between variables and the role that geography plays in those relationships. For PEPFAR programming, the models could show relationships between social factors and HIV infections and treatment. A disadvantage is its dependence on solid data for the factors to be assessed. Additionally, the resulting data products may require detailed explanation.Wabiri, Shisana, Zuma, Freeman, Spatial and Spatio-temporal Epidemiology 16 (2016).

Targeting elimination of mother-to-child HIV transmission efforts using geospatial analysis of mother-to-child transmission in Zimbabwe – Analyzed serosurvey data, aggregated individual data to catchment areas and estimated mother-to-child transmission. The advantage is its simplicity. Aggregating data and mapping it is easy with any number of geospatial tools. The resulting maps are effective data products that can show areas of need. The disadvantage is that it is simplistic and may not consider the complexity of populations seeking care. McCoy, Fahey, Buzdugan, Mushavi, Mahomva, Padian, Cowan, AIDS 2016, 30:1829-1837.

John Spencer is Senior GIS Technical Specialist at MEASURE Evaluation.

Leave a Comment

Your email address will not be published. Required fields are marked *