Analyses were restricted to children who had completed PAH
metabolite and cockroach IgE levels measured at both 5 and
9 years, with a final sample size of 188 (Fig. 1). Proportion was
calculated for categorical characteristics of the samples included
and excluded from the current analysis. Chi-square test was used
to detect difference in the proportions between the two samples.
Spearman correlation coefficient was calculated for correlation in
PAH metabolites between measures at ages 5 and 9–10 while
McNemar test was used to detect differences between paired
proportions at the two ages. The change in proportion of variables
of interest from age 5 to age 9–10 was examined by a model for
binary repeated measures. To obtain a measure for exposure pattern
in each individual PAH compound, we first dichotomized the
SG adjusted urinary PAH metabolites (ages 5 and 9–10) at their
median values of age 5 measures. Then we categorized the repeated
measures of PAH metabolites for each child into four
groups (age 5-age 9–10: High–High [repeatedly high], High–Low,
Low–High vs. Low–Low [reference]) for each individual PAH
compound. To assess the associations between the composed
variables for the repeated PAH measures at ages 5 and 9–10 and
allergic sensitization at age 9, we used a modified Poisson regression
for the dichotomous outcome (yes vs. no cockroach
sensitization at age 9), with and without controlling for potential
confounding factors of sex, race/ethnicity, maternal asthma, prenatal
ETS exposure, cockroach allergen exposure (either at age 5 or
9) and cockroach sensitization at age 5. To aid interpretation of
result, we derived relative risk (RR) along with 95% confidence
interval (CI) from the estimated models fit to the data.
Data analysis was conducted with SPSS version 22.0 (SPSS Inc.,
Chicago, IL, USA). All tests were two-sided with significance level
of 0.05.
3. Results
3.1. Cohort characteristics
Table 1 shows that children included in anal