Spatial Clustering of Infectious Syphilis in Victoria, Australia: Disentangling MSM Residential Geography from Epidemiological Heterogeneity

Background: Victoria, Australia, has experienced a sustained increase in infectious syphilis driven predominantly by transmission among men who have sex with men (MSM). We applied spatial statistical methods to quantify geographic clustering, identify area-level correlates, and disentangle demographic concentration from epidemiological heterogeneity using Census-derived MSM population estimates. Methods: Publicly available notification data for 75 Victorian local government areas (LGAs), 2019-2024, were analysed. Global/Local Moran's I and Getis-Ord Gi* identified spatial clusters. Spatial regression and geographically weighted regression compared Census same-sex male couple density against a demographic proxy (proportion males aged 20-44). LGA-level MSM populations, estimated using Wilson and Shalley's Census-based method, were validated against three external benchmarks. Standardised morbidity ratios (SMRs) with Poisson confidence intervals quantified residual per-MSM rate heterogeneity. Results: Strong spatial clustering was identified (Moran's I = 0.560, p < 0.001), with seven of 11 LISA high-high clusters persisting across all six study years. Same-sex couple density outperformed the demographic proxy (R-squared = 0.664 vs 0.530; AIC improvement 24.8 units). MSM population adjustment reduced spatial autocorrelation by 74.5% and halved the coefficient of variation of notification rates, though significant residual heterogeneity persisted (Poisson chi-squared p < 0.001). SMR analysis revealed that the highest per-MSM rates were in outer suburban LGAs (Melton 3.25, Frankston 2.06, Brimbank 1.87), while several inner Melbourne hotspot LGAs had SMRs below 1.0. Negative binomial regression with MSM offset identified socioeconomic disadvantage (IRR = 0.995, p < 0.001) and remoteness (IRR = 0.41, p < 0.001) as predictors of per-MSM rates. Conclusions: Three-quarters of Victoria's apparent syphilis clustering reflects MSM residential geography. Residual per-MSM heterogeneity reveals an outer suburban excess that challenges resource allocation based on raw notification counts and supports decentralised testing expansion.