Exposure was assigned based on the place of residence of each
participant, with Launceston residents classed as ‘exposed’ and
Hobart residents classed as ‘unexposed.’ The characteristics of the
exposed and unexposed populations were compared using chisquared
tests. Symptom prevalence in Launceston was compared to
symptom prevalence in Hobart using logistic regression modelling,
with each symptom examined separately as a dependent variable. The
models reported odds ratios (OR) with 95% confidence intervals (95%
CI), and included possible confounders as independent variables. The
baseline model included gender as a co-variate, but age was not
considered because all participants were, by definition, the same age
(mean age 43.0 years, ±0.06) and the inclusion of age as an
independent variable in the regression models did not alter the
outcomes. Additional potential confounders were systematically
added to the model, and those that did not alter the outcome by
10% or more were removed. Interaction factors were used in the
model to adjust for effect modification between co-variates.
Exposure was assigned based on the place of residence of each
participant, with Launceston residents classed as ‘exposed’ and
Hobart residents classed as ‘unexposed.’ The characteristics of the
exposed and unexposed populations were compared using chisquared
tests. Symptom prevalence in Launceston was compared to
symptom prevalence in Hobart using logistic regression modelling,
with each symptom examined separately as a dependent variable. The
models reported odds ratios (OR) with 95% confidence intervals (95%
CI), and included possible confounders as independent variables. The
baseline model included gender as a co-variate, but age was not
considered because all participants were, by definition, the same age
(mean age 43.0 years, ±0.06) and the inclusion of age as an
independent variable in the regression models did not alter the
outcomes. Additional potential confounders were systematically
added to the model, and those that did not alter the outcome by
10% or more were removed. Interaction factors were used in the
model to adjust for effect modification between co-variates.
การแปล กรุณารอสักครู่..