Visualizing the Geography of HIV Observational Cohorts With Density-Adjusted Cartograms

Abstract

Background: Maps are potent tools for describing the spatial distribution of population and disease characteristics and, thereby, for appropriately targeting public health interventions. People with HIV (PWH) tend to live in densely populated and spatially compact areas that may be difficult to visualize on maps using unadjusted geographic or political borders.

Setting: To illustrate these challenges, we used geographic data from adult PWH at the Vanderbilt Comprehensive Care Clinic (VCCC) in Nashville, Tennessee, and aggregated data from the North American AIDS Cohort Collaboration on Research and Design (NA-ACCORD) from 1998 to 2015.

Methods: We compared choropleth maps that use differential shading of political/geographic boundaries with density-adjusted cartograms that allow for shading and deformed boundaries according to a variable of interest, such as PWH.

Results: Cartograms enlarged high-burden areas and shrank low-burden areas of PWH, improving visual interpretation of where to focus HIV prevention and mitigation efforts, when compared with choropleth maps. Cartograms may also demonstrate cohort representativeness of underlying populations (eg, Tennessee for VCCC or the United States for NA-ACCORD), which can guide efforts to assess external validity and improve generalizability.

Conclusion: Choropleth maps and cartograms offer powerful visual evidence of the geographic distribution of HIV disease and cohort representation and should be used to guide targeted public health interventions.

Figure 1.. Tennessee county choropleth map, continuous, and Doling cartogram.

Choropleth map of Tennessee counties (panel A) colored by Vanderbilt Comprehensive Care Center (VCCC) HIV patients in care. Continuous (panel B) and Dorling Cartogram (panel C) of the cumulative VCCC cohort density-adjusted by the total HIV+ prevalence in each Tennessee county in 2015. Counties are labelled to aid in interpretation. “Cum.” abbreviates “Cumulative” and includes people living with HIV in care at the VCCC from 1998–2015.

Figure 2.. Dorling cartogram of contiguous US counties.

Dorling cartogram of the cumulative cohort of the Vanderbilt Comprehensive Care Center (VCCC) density-adjusted by the total HIV+ prevalence by US county, with darker blue shading indicating counties with VCCC cohort participants. Circle outline colors are the same for counties within the same state and vary between states. Counties above the 97.5 percentile for prevalent HIV cases are labeled with their state’s two-letter abbreviation. Ending the HIV Epidemic phase 1 priority jurisdictions have thicker county borders. “Cum.” abbreviates “Cumulative” and includes people living with HIV in care at the VCCC from 1998–2015.

Figure 3.. Choropleth maps, continuous cartograms, and Dorling cartograms of contiguous US states.

Panels A and B show choropleth maps of the prevalent population of people living with HIV and HIV prevalence proportion (HIV prevalence per 100,000 population) in 2015 respectively. Panels C-F show the NA-ACCORD population with political borders smoothly deformed with continuous transformations (C and D) or non-continuous density-adjusted with Dorling transformations (E and F) according to underlying populations living with HIV (C and E) or new HIV diagnosese (D and F) in 2015. “Cum.” abbreviates “Cumulative” and includes people living with HIV in care at the NA-ACCORD site from 1998–2015 and “Dx” abbreviates “Diagnoses.”