Figure 7

Joinpoint regression models fitted to proportions of prostate cancer cases diagnosed late in several Florida counties. Joinpoint regression was conducted on the time series of noise-filtered proportions of late-stage diagnosis recorded in six heavily populated counties and the two groups of rural counties identified in Figure 1A. Within each county or group of counties, the hypothesis of parallelism of models for white and black males was rejected at α = 0.05 and the disparity statistic B _rr' was computed.