Article Text

Download PDFPDF

P3.243 Influence of Scale and Zone on Syphilis Trend Interpretation
  1. V Escamilla1,2,
  2. K Hampton3,
  3. P A Leone3,4,1,
  4. W C Miller1,3
  1. 1Division of Infectious Diseases, Department of Medicine, University of North Carolina, Chapel Hill, NC, United States
  2. 2Carolina Population Center, University of North Carolina, Chapel Hill, NC, United States
  3. 3Department of Epidemiology, Gillings School of Global Public Health, University of North Carolina, Chapel Hill, NC, United States
  4. 4HIV/STD Prevention and Care Branch, North Carolina Division of Public Health, Raleigh, NC, United States


Background We investigated spatial clustering of primary and secondary syphilis rates in North Carolina (2003–2010) using multiple scales and geographical boundaries. We examined the influence of changes in scale and boundary on identification of spatial clusters using two cluster detection methods, local Moran’s I and Kulldorff’s space scan statistic.

Methods We used two cluster detection methods: (1) local Moran’s I with empirical Bayes (EB) standardised rate statistic and (2) Kulldorff’s space scan statistic using a variable size moving circular window. We evaluated three geographic zones with decreasing boundary area, North Carolina, Piedmont region, and Mecklenburg County, at two spatial scales (census tract, census block group). We report results for Mecklenburg County.

Results Using local Moran’s I, block group clusters were in the same location as tract clusters, but were more concentrated. Median rates were higher among block group clusters compared with tract clusters. As boundary areas decreased, some clusters in peripheral tracts and block groups were lost. High rate block groups were more likely to persist, while some high rate peripheral tracts were lost.

With Kuldorff’s scan statistic, block group clusters were more concentrated than clusters in census tracts. Reducing boundary areas had little effect on census tract clusters detected using Kuldorff’s scan statistic. Cluster size decreased significantly when the boundary was restricted to Mecklenburg County. The reduction in cluster size reflected loss of a few high rate block groups peripherally and many block groups with a rate of zero.

Clusters detected using local Moran’s I and Kuldorff’s scan statistic overlapped, but Kuldorff’s scan clusters were much larger with a high proportion of zero rate tracts/block groups.

Conclusion In efforts to understand STI epidemiology spatially, investigators must carefully consider the spatial scale, geographical area of interest, and cluster detection approach.

  • scale
  • spatial epidemiology
  • Syphilis

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.