Zulu, L. C., Kalipeni, E., & Johannes, E. (2014). Analyzing spatial clustering and the spatiotemporal nature and trends of HIV/AIDS prevalence using GIS: The case of Malawi, 1994-2010. BMC Infectious Diseases, 14(1), 285-285. doi:10.1186/1471-2334-14-285

 

This study offers insights into the spatiotemporal trends of HIV prevalence at the national, regional, district, and sub-district scales in Malawi from 1994 to 2010, as well as the factors associated with HIV prevalence for 2010.

The authors first turned to the literature on the use of GIS and spatial analysis in analyzing HIV prevalence and identifying common factors impacting HIV prevalence. Then, they conducted their own analysis using GIS, spatial statistical models such as Moran’s I, and descriptive statistics such as median HIV prevalence. GIS plays a vital role in this analysis and an array of spatial analysis tools were used. Inverse distance weighting was first used to interpolate and generate continuous surfaces of HIV prevalence from point data obtained from the surveillance antenatal clinics (ANC). This allows observations and predictions to be made even with a small sample size. Then, HIV hotspots and cold spots were identified using spatial autocorrelation measures such as the Anselin Local Moran’s I. Finally, correlation and regression analyses (ordinary least square model) were used to discern the factors associated with HIV prevalence, and compare their spatial clustering with HIV clustering.

The main argument presented in this article is that Malawi needs to develop spatial targeting HIV interventions and policies as there are multiple geographically differentiated HIV/AIDS epidemics, rather than a single one. The authors establish evidence for this spatial observation using Anselin Local Moran’s I. They found a noticeable pattern of high HIV-prevalence districts next to high ones (i.e. HIV ‘hotspots’) clustering in the highly urbanized southern region, while a low HIV prevalence in the Northern and Central regions. This local variation disrupts the broad assumption of urban/rural differences presented in other literature. In light of that, the authors argue that the existing one-size-fits-all interventions and policies are highly ineffective in addressing HIV in Malawi. Furthermore, local epidemics are shaped by different configurations of factors. When examining the results of the regression analysis, the authors found that the four explanatory factors—mean distance to main roads, mean travel time to nearest transport, the percentage of population that had taken an HIV test ever, and percentage attaining a senior primary education—have different impacts on the locations. Thus, spatial specific intervention is required.

There are two main limitations in this study. First, the sample size of data collected from ANC is very small—only 19 HIV prevalence data point are available to conduct the analysis for the entire country. Although the authors tried to address this issue by using interpolation, the result is not entirely representative and may be inaccurate at a large scale. Second, the OLS model assumes that processes occur in isolation with space. As the purpose of this study is to look at the local spatiotemporal variation in HIV prevalence, geographically-weighted regression model which takes into account spatial heterogeneity would be more suitable for this analysis. Overall, this study is well-structured and comprehensive in that the implementation and rationale behind each step are well-justified and explained in great detail. It provides excellent foundations on which further research can build to establish longitudinal analyses of HIV at the multiple scales in another country.

Rating: 8.5/10

---

Tara’s presentation on healthfulness index really stood out to me this week.

Sadler, R. C., Hippensteel, C., Nelson, V., Greene-Moton, E., & Furr-Holden, C. D. (2019). Community-engaged development of a GIS-based healthfulness index to shape health equity solutions. Social Science & Medicine, 227, 63-75. doi:10.1016/j.socscimed.2018.07.030

The logic behind the healthfulness index is that there is a correlation between health outcomes and social determinants of health. Members of the communities in Flint, Michigan were asked to determine what variables promoted or hindered healthy behaviours using the analytic hierarchy process. Comparing community inputs and academic inputs, the authors found that community members generally think that social issues have a higher influence than experts. This article is an excellent example of analysis based on subjective values. The approach they used also encourages public participation in GIS analysis.