Dose, Timing, and Spectrum of Prenatal Antibiotic Exposure and Risk of Childhood Asthma

Abstract

Background: The potential for prenatal antibiotic exposure to influence asthma risk is not clear. We aimed to determine the effect of timing, dose, and spectrum of prenatal antibiotic exposure on the risk of childhood asthma.

Methods: We conducted a population-based cohort study of 84 214 mother-child dyads to examine the association of prenatal antibiotic exposure and childhood asthma using multivariable logistic regression models.

Results: Sixty-four percent of pregnant women received antibiotics. Prenatal antibiotic exposure was associated dose-dependently with increased odds of childhood asthma (adjusted odds ratio [aOR] for interquartile increase of 2 courses [interquartile range, 0-2], 1.26 [95% confidence interval {CI}, 1.20-1.33]). Among children exposed to at least 1 course in utero, the effect of timing at the first course was moderated by total maternal courses. Among pregnant women receiving a single antibiotic course, timing of exposure had no effect on childhood asthma risk. Among women receiving > 1 course, early exposure of the first course was associated with greater childhood asthma risk. Compared to narrow spectrum-only antibiotic use, broad spectrum-only antibiotic exposure was associated with increased odds of asthma (aOR, 1.14 [95% CI, 1.05-1.24]). There were effect modifications (P < .001) by maternal asthma on total courses, and on timing of the first course, significant only among those without maternal asthma.

Conclusions: Increased cumulative dose, early pregnancy first course, and broad-spectrum antibiotic exposure were associated with childhood asthma risk. Our study provides important evidence supporting judicious prenatal antibiotic use, particularly timing of use and choice of antibiotics, in preventing subsequent childhood asthma.

Keywords: prenatal antibiotic exposure; antibiotic course timing; childhood asthma; dose-response relationship; genetic predisposition.

Figure 1.

Relationship between the cumulative number of prenatal antibiotic courses and childhood asthma. Multivariable logistic regression adjusted for maternal age, asthma, education, smoking, presence of group B Streptococcus infection, delivery method, and infant characteristics including birth hospitalization length of stay, gestational age, birth weight, infant race, sex, and number of living siblings at home. Gray band around the line is the 95% confidence interval.

Figure 2.

The association of timing at first prenatal antibiotic exposure with childhood asthma among children exposed to at least 1 course of prenatal antibiotics. Multivariable logistic regression adjusted for maternal age, asthma, education, smoking, presence of group B Streptococcus infection, delivery method, and infant characteristics including birth hospitalization length of stay, gestational age, birth weight, infant race, sex, and number of living siblings at home. Gray bands around the line represent the 95% confidence interval.

Figure 3.

Effect modification of the timing of the first prenatal antibiotic exposure (gestational age in days) and the cumulative number of prenatal antibiotic courses on risk of childhood asthma. Multivariable logistic regression adjusted for maternal age, asthma, education, smoking, presence of group B Streptococcus infection, delivery method, and infant characteristics including birth hospitalization length of stay, gestational age, birth weight, infant race, sex, and number of living siblings at home. The subjects with ≥ 5 courses of antibiotics have wide confidence bands due to sparse data at later gestational ages. In the group with ≥ 5 courses of prenatal antibiotic exposure, the gestational age of first exposure tended to be earlier with a median of 25 (interquartile range [IQR], 5–51) days, whereas in those with 1–4 courses of antibiotics, median gestational age of first exposure was 95 (IQR, 43–167) days.

Figure 4.

Effect modification of maternal asthma status and the cumulative number of prenatal antibiotic courses on the risk of childhood asthma. Multivariable logistic regression adjusted for maternal age, education, smoking, presence of group B Streptococcus infection, delivery method, and infant characteristics including birth hospitalization length of stay, gestational age, birth weight, infant race, sex, and number of living siblings at home. The cumulative number of prenatal antibiotic courses was assessed with restricted cubic splines (4 knots). The gray band around the lines indicates the 95% confidence interval.

Figure 5.

Effect modification of maternal asthma status and timing of the first prenatal antibiotic course on the risk of childhood asthma among children exposed to at least 1 course of prenatal antibiotics. Multivariable logistic regression adjusted for maternal age, education, smoking, presence of group B Streptococcus infection, delivery method, and infant characteristics including birth hospitalization length of stay, gestational age, birth weight, infant race, sex, and number of living siblings at home. Gestational age in days at the first course was assessed with restricted cubic splines (3 knots). The gray band around the lines indicates the 95% confidence interval.

Figure 6.

Diagram depicting the hypothesized causal mechanisms leading to the observed association between prenatal antibiotic exposure and subsequent childhood asthma development.