RR for persons living in highly exposed census blocks
compared with those living in the slightly exposed block
groups (90th percentile/2.5th percentile) was 1.120 (95%
CI 1.002 – 1.251).
Subgroup analyses per gender are reported in Tables 3 and
4. Unsurprisingly, best fitting models were different.
Among males, modelled ground-level dioxin concentrations
revealed significant in the univariate model (p-value
= 0.04), but not in the multivariate approach (Table 3).
Conversely, RRs appeared statistically significant among
females, both in the univariate (p-value = 10-5) and the
multivariate (p-value = 0.04) models, yielding a RR of
1.178 (95% CI 1.013 – 1.369) for the latter.
No extra Poisson variation was highlighted with the hierarchical
Bayesian Poisson regression for any model
(results not shown).