Results
The analytical sample included 879 boys and 970 girls. By age seven
years, 37.6% of children experienced one or more wheezing episodes
(Table 1). EOT wheeze was the most common phenotype (20.5%); followed
by LO (9.3%) and EOP (7.8%). Paediatrician-diagnosed LRTIs were common
(46.8%). Every fifth mother reported smoking during pregnancy (19.6%)
and 30.8% reported that their toddler had some exposure to tobacco
smoke. Majority of children (80.8%) were breastfed; over a half
(53.5%) for less than four months.
Wheezing was linked to smoking during pregnancy (p=0.044), child’s SSE
(p=0.008), LRTI (p<0.001), and lack/short duration of
breastfeeding (p=0.035). Wheezing was more common among boys
(p<0.001), Brno residents (p=0.031), second-born and higher
birth-order children (p=0.044), and children of mothers with a history
of asthma (p=0.028).
In bivariate models showing unadjusted relationships, the relative risk
of wheezing vs. no wheezing was 54%-79% higher among children with
high vs. low exposure to prenatal life events (Table 2). Relative
risk ratios (RRR) for wheeze phenotypes were 1.54 (95% CI 1.16-2.05,
p=0.003) for EOT; 1.72 (95% CI 1.13-2.63, p=0.012) for EOP; and 1.79
(95% CI 1.22-2.61, p=0.003) for LO. Increases in the relative risk of
wheeze linked to postnatal life events were even higher. High vs.
low exposure was associated with 57%-145% increases (EOT, RRR=2.11,
95% CI 1.58-2.82, p<0.001; EOP, RRR=2.45, 95% CI 1.55-3.90,
p<0.001; LO, RRR=1.57, 95% CI 1.06-2.32, p=0.026).
Multivariate Results
Adjusted Relationships between Perinatal Life Events and Wheeze.In Table 2, statistically significant differences in the likelihood of
several wheeze phenotypes by high vs. low exposure to life events remain
evident even after adjustment for covariates. Children with highprenatal exposure to life events have a 44% higher relative risk
of EOT wheeze compared to counterparts with low exposure (RRR=1.44, 95%
CI 1.06-1.95, p=0.020). Moreover, they have a 69% higher relative risk
of LO wheeze (RRR=1.69, 95% CI 1.13-2.53, p=0.010). EOP wheeze
approaches significance with RRR of 1.56 (95% CI 0.99-2.44, p=0.053).
For postnatal life events, high vs. low exposure is related to a
74% increase in relative risk of EOT wheeze (RRR=1.74, 95% CI
1.28-2.36, p<0.001); the corresponding increase for EOP wheeze
is 101% (RRR=2.01, 95% CI 1.23-3.26, p=0.005). These results support
H1 and H2.
Figure 2 illustrates predicted probabilities of each wheeze phenotype
based on marginal effects calculated for four categories of stress
exposure, overall low stress (low prenatal/low postnatal life
events); increasing stress (low prenatal/high postnatal life
events); decreasing stress (high prenatal/low postnatal life
events); and overall high stress (high prenatal/high postnatal
life events). In all stress categories, the majority of children never
wheeze, but the probability of staying wheeze-free throughout the
observation period is considerably lower among children with overall
high stress (51.9%) vs. children with overall low stress (72.2%). The
predicted probability of EOT wheeze more than doubles among children in
the overall high stress category (9.4%) compared to overall low stress
category (4.5%) Predicted probabilities corresponding to overall low
stress vs. overall high stress are 15.4% vs. 26.2%, respectively, for
EOT wheeze and 8.0% vs. 12.5%, respectively, for LO wheeze.
Mediation between Prenatal Life Events and Wheeze. Models of
mediators are displayed in Table 3. Unstandardized coefficients are
reported for all types of models to facilitate interpretation. Higher
prenatal life events predict higher postnatal life events (high vs. low
prenatal life events, unstandardized coefficient b=0.77, 95% CI
0.49-1.06, p<0.001). Decomposition analysis reveals that
postnatal life events partially mediate between prenatal life events on
any wheeze vs. no wheeze (high vs. low prenatal events, direct effect
odds ratio [OR]=1.51, 95% CI 1.18-1.93, p=0.001; indirect effect
OR=1.13, 95% CI 1.06-1.21, p<0.001; total effect OR=1.71 95%
CI 1.34-2.18, p<0.001), supporting H3a. Prenatal life events
are unrelated to smoking during pregnancy, SSE, breastfeeding, LTRI, and
allergy, ruling out mediation through these factors.
Mediation between Postnatal Life Events and Wheeze. Children with
higher exposure to postnatal life events are more commonly diagnosed
with LRTI (Table 2) and LRTI mediates between postnatal life events and
any wheeze (high vs. low postnatal events, direct effect OR=1.85, 95%
CI 0.05-74.31, p=0.745; indirect effect OR=1.06, 95% CI 1.02-1.09,
p=0.002; total effect OR 1.95, 95% CI 0.05-78.75, p=0.723). This result
supports H4c. Postnatal life events are also linked to SSE, another
significant mediator (direct effect OR=1.86, 95% CI 0.04-82.06,
p=0.749; indirect effect OR=1.02, 95% CI 1.001-1.05, p=0.035).
Mediation through breastfeeding is unsupported.
To illustrate multivariate results, Figure 3 displays estimated
statistical effects for each pathway in the proposed conceptual model.
It demonstrates the mediation of prenatal stress through postnatal
stress, mediation of postnatal stress through respiratory infections,
and mediation of postnatal stress through SSE.
Covariate Effects (results upon request). In the final
multinomial logistic regression model of wheeze phenotypes, boys are
more likely to develop early-onset wheeze compared to girls (EOT,
RRR=1.60; 95% CI 1.25-2.04, p<.001; EOP, RRR=1.61; CI
1.13-2.31, p=0.009). EOP wheeze is related to maternal asthma history
(RRR=3.08, 95% CI 1.44-6.58, p=0.004).