Tick-borne relapsing fever (TBRF) is a rare bacterial disease caused primarily by Borrelia hermsii and Borrelia turicatae in the western United States and transmitted by Ornithodoros species soft ticks. No spatial analyses have been attempted for TBRF, and previous epidemiologic studies were limited to case series and outbreak investigations. This study employed ArcGIS to map counties and zip codes with identified cases of TBRF and neighboring control counties and zip codes. A total of 140 counties with reported cases of TBRF, identified in a previous publication, and 243 counties with no reported cases in 12 states were included in the county level analysis. The zip code level analysis included 60 zip codes with cases of TBRF and 193 control zip codes in California and Washington, using information provided by state health departments. Ecologic factors, including elevation, precipitation, average minimum temperature, average maximum temperature, and land cover, in these areas were compared by frequency analysis and logistic regression analyses. The occurrence of TBRF was associated with elevation, temperature, and evergreen forest land cover in county level analyses, and with elevation and temperature in zip code level analyses. No associations were found with precipitation or additional land cover variables and TBRF occurrence. Counties (0.25 ≥ p ≥ 0.0003) and zip codes (0.0007 ≥ p ≥ 0.03) with cases were seen in higher proportions at elevations above 500 meters than control counties and zip codes, and elevation was included in logistic regression models at both levels of analysis. A higher proportion of counties with cases were observed in the middle of the range of temperature values, while control counties were evenly distributed (0.01 ≥ p ≥ 0.0004). The association with temperature at the zip code level was less consistent, with higher case zip code proportions observed at lower temperatures (0.08 ≥ p ≥ 0.01). A temperature variable was included in logistic regression analyses at both levels of analysis. Evergreen forest was the majority land cover type in a greater proportion of counties with cases when compared to control counties (total land cover p = 0.04) and this variable was only significant in the county level logistic regression analyses. The distribution of land cover variables was not significant at the zip code level (p = 0.82) and no zip code level land cover variables were significant in logistic regression analyses.

Similar associations were observed when using logistic regression to analyze high risk counties and control counties (p = 0.005), and high risk zip codes and control zip codes (p = 0.006). Zip code level analyses of California produced a logistic regression model containing an elevation variable (p = 0.0002), while the best model for Washington contained the same variables found in the complete zip code level analysis (p = 0.07). These results suggest that ecologic factors including elevation and temperature play a role in areas where TBRF occurs. These factors likely influence the distribution and/or abundance of the tick vectors responsible for this disease or their preferred hosts. Further refinement of these analyses could lead to the construction of a predictive model that could be used to highlight areas of increased risk of TBRF.