How Place Shapes Assortativity: Sexual Partnerships and Race in
Toomey, Christiana Elizabeth
Master's Thesis (74 pages)
Committee Chair / Thesis Adviser: Sanchez, Travis
Research Fields: Health Sciences, Public Health
Partnering Agencies: Does not apply (no collaborating organization)
Keywords: racial assortativity; MSM; sexual-network measurement; online questionnaire; geospatial analysis
Program: Rollins School of Public Health, Epidemiology (Epidemiology)
Background: Men who have sex with men (MSM) remain the group most affected by the United States' HIV/AIDS epidemic. Racial sexual assortativity, or the degree to which a man chooses partners from his same race, may play a role in transmission risk, especially in light of recent findings regarding concurrency and partner-to-partner risk among MSM. While existing methods have been used to measure risk conferred by concurrency, and this risk has been compared by race (on individual-, individual-to-partner, and partner-to-partner levels), the relationship between men in space, and their dyadic racial sexual assortativity, has not.
Objective: To evaluate the application of a geostatistical analysis method as a means for quantifying clustering of similar behavior, as demonstrated by spatial clustering of similar dyadic racial sexual assortativity, in an online study of MSM.
Methods: Data collected from participants in the June 2010-December 2012 prospective observational study of Atlanta MSM were collapsed to census tract. A new method was used to calculate racial assortativity prevalence per tract, in which participants were classified as either fully-assortive (only choosing male sexual partners within their stated race) or non-assortive (choosing some or all male sexual partners outside their race). Geospatial statistics were generated on the full group of participant census tracts, as well as strata for tracts in which black and white participants resided. The summary statistic Moran's I was calculated for each group (overall, race strata) to determine significance in clustering (at p<0.001), and the Getis-Ord Gi* statistic was calculated at each census tract centroid.
Results: In the analysis of 349 census tracts, the z-score for spatial autocorrelation (Moran's I) was 3.5 (p=0.0004) for the fully-assortive prevalence. In the analysis of 238 census tracts containing black participants, the z-score was -13.4 (p<0.0001) for fully-assortive prevalence. In the analysis of 177 census tracts containing white participants, the z-score was 2.9 (p=0.004) for fully-assortive prevalence. Spatial clustering maps revealed one statistically significant hotspot of fully-assortive tracts southeast for the overall sample and central and southwest Atlanta for black MSM.
Conclusions : Geographic differences in assortativity highlight an area of further exploration of root causes in racial differences in HIV risk in MSM.
Table of Contents
Table of Contents
List of Tables
Table 1. Involvement Participant Characteristics at Baselineβ¦β¦β¦β¦β¦...β¦β¦β¦32
Table 2. Distribution of Fully Assortive Partnership Prevalence Per Reporting Census Tract, Black Participantsβ¦β¦β¦β¦β¦β¦..β¦β¦β¦β¦β¦β¦β¦β¦β¦β¦β¦β¦β¦β¦.33
Table 3. Distribution of Fully Assortive Partnership Prevalence Per Reporting Census Tract, White Participantsβ¦β¦β¦β¦β¦β¦β¦β¦β¦β¦β¦β¦β¦β¦β¦β¦β¦β¦β¦β¦..34
List of Figures
Figure 1. Participants Overall - Hot spot map of fully-assortive partnership..β¦.35
Figure 2. Black Participants- Hot spot map of fully-assortive partnershipβ¦...β¦36
Figure 3. White Participants- Hot spot map of fully-assortive partnershipβ¦β¦..37
Appendix 1. P-values for Computed Assortativity Variable Across Atlanta Census Tract, All Participantsβ¦β¦β¦β¦β¦β¦β¦β¦β¦β¦β¦β¦β¦β¦β¦β¦β¦β¦β¦β¦β¦...38
Appendix 2. P-values for Computed Assortativity Variable Across Atlanta Census Tract, Black Participantsβ¦β¦β¦β¦β¦β¦β¦β¦β¦β¦β¦β¦β¦β¦β¦β¦β¦β¦β¦.......47
Appendix 3. P-values for Computed Assortativity Variable Across Atlanta Census Tract, White Participantsβ¦β¦β¦β¦β¦β¦β¦β¦β¦β¦β¦β¦β¦β¦β¦β¦β¦β¦β¦β¦..53
Appendix 4. SAS Codeβ¦β¦β¦β¦β¦β¦β¦β¦β¦β¦β¦β¦β¦β¦β¦β¦β¦β¦β¦β¦β¦β¦β¦β¦...58