Table 2.

Final model parameter estimates for negative binomial regression models.

Model	Parameter estimates
Country-wide analysis	Intercept	Urban (20)	Forest (20)	Mean elevation	Wetness index standard deviation	Random district effect (${\sigma }_d^2$)
	0.76 (1.71)	-3.62 (1.96)	1.68 (0.88)	-0.85 (0.22)	3.06 (1.27)	0.55 (0.37)
Cluster 1	Intercept	Urban (0.5)
	-6.05 (0.14)	-5.28 (1.81)

Response variable is the number of Buruli ulcer cases in villages in Benin, West Africa. Models for two analyses are presented: a country-wide analysis (n = 327 villages), and an analysis for Buruli ulcer disease in cluster 1 (see methods; n = 100 villages). Cluster 1 had greater than expected prevalence (see Figure 3 for location of cluster). Parameter estimates are followed by standard errors in parentheses. Land use/cover types are percent in a buffer surrounding each village, the buffer width (km) follows the land use/cover type in parentheses.