Introduction

Asthma imposes a great health burden globally. It was estimated that about 300 million people were affected in 2011[1], and 250,000–345,000 people die from asthma each year [2]. From public health perspectives, more attention is needed to the etiology of asthma, especially in children, for whom the incidence of asthma was more than three times of that in adults [3].

Previous studies showed that maternal antibiotics exposure during pregnancy was associated with an increased risk of childhood asthma but the results were inconsistent. [4–9]. For example, Stensballe et al. reported that maternal exposure to different types of antibiotics in the third trimester was associated with elevated childhood asthma incidence [8], but McKeever et al. [7] failed to find any significant associations. Moreover, none of the previous studies examined the periconceptional window, which is a crucial period for embryo development [10]. Therefore, we conducted the present study to assess the associations between maternal exposure to different types of antibiotics before and during pregnancy and childhood asthma in the Collaborative Perinatal Project (CPP), a large prospective birth cohort study in the United States in 1959–1976.

Methods

Study population

A detailed description of the study population and method has been provided elsewhere [11]. Briefly, 46,021 women with 56,990 pregnancies were enrolled at 12 sites in the United States from 1959 to 1965. These women were routinely interviewed during pregnancy and the newborns were followed up. Approximately 80 percent of the children were followed up to the age of seven to eight years. Physical examinations of neurological, neurosensory, cognitive development, general health and physical growth were taken at eight months, one, three, four, seven, and eight years of age in children [12]. Among 53,647 singleton births, 40,063 were followed until at least 7 years of age. The CPP was carried out by researchers who conducted the original study from 1959–1976.We excluded 156 children without a record of maternal periconceptional medication, leaving 39,907 children for the final analysis (Fig 1).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Study population flow chart, Collaborative Perinatal Project, United States, 1959–1965.

https://doi.org/10.1371/journal.pone.0140443.g001

Results

The prevalence of asthma was 5.5% (2,201/39,907) in this population. Baseline characteristics of these children are shown by asthma status in Table 1. No significant differences were observed in maternal age at pregnancy, maternal educational level, maternal smoking during pregnancy, and delivery mode between asthma and non-asthma groups. However, the asthmatic children had a higher proportion of unmarried, asthmatic mothers, mothers with drug allergy history, male infants and black race (all p<0.05).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Maternal and Child Characteristics by Asthma Status in the Collaborative Perinatal Project, United States, 1959–1965.

https://doi.org/10.1371/journal.pone.0140443.t001

A total of 10,534 (26.4%) mothers were treated with antibiotics one month prior to LMP or during pregnancy. The vast majority of them used one type of antibiotics (22.5%). Penicillin was the most commonly used antibiotic (15.6% of all mothers) followed by sulfonamides and trimethoprim (10.1% of all mothers). Most mothers received antibiotic treatment in the second (12.8% of all mothers) or third (12.7% of all mothers) trimesters.

Table 2 shows the associations between maternal exposure to various antibiotics before and during pregnancy and asthma in the offspring. Maternal use of penicillin or chloramphenicol was associated with a significantly increased risk of childhood asthma after adjusting for potential confounders (adjusted OR = 1.21 [95% CI1.08–1.36] for penicillins; adjusted OR = 1.72 [95% CI1.14–2.59] for chloramphenicols). We further separated the results by the time of medication. Penicillins or chloramphenicols use in first trimester was significantly associated with childhood asthma (adjusted OR = 1.09 [1.04–1.13] and1.23 [1.01–1.51], respectively).When we further controlled for low birthweight and preterm birth as potential confounders in the multilevel logistic regression models, the results didn’t change.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Adjusted and unadjusted risk of childhood asthma exposed to maternal antibiotic use during 4 weeks before LMP and pregnancy and followed to 7 years of age.

https://doi.org/10.1371/journal.pone.0140443.t002

A sensitivity analysis was conducted by excluding women who reported having exposed to two or more types of antibiotics before or during pregnancy. The results remained virtually the same (not shown). When we stratified the results by the time of medication, again the results were very similar (S1 Table).

Discussion

In this prospective cohort study, we found that use of penicillins or chloramphenicols during early pregnancy was associated with an increased risk of asthma in offspring after controlling for potential confounders. These findings suggest that early life exposure to certain antibiotics might increase the risk of childhood asthma.

Our finding is consistent with two Danish longitudinal cohort studies, in which if mothers used antibiotics at any time during pregnancy, the children had an increased risk of asthma hospitalization (hazard ratio 1.17 [1.00–1.36]), and inhaled corticosteroids (1.18 [1.10–1.27]) [8]. However, a recent Swedish study using sibling analysis suggested that previous positive findings may have been due to confounding by indication or uncontrolled genetic and environmental confounders [18]. Although the sibling analysis may have adjusted residual familial factors better, the exposure combined all types of antibiotics in that study [18]. Our study showed when all antibiotics were combined, the significant association found in penicillin was reduced, indicating that the association may be antibiotic-specific.

Some old antibiotics are no longer being used in contemporary practice. However, our findings raised several questions that may still have important clinical and pharmacodevelopmental implications. In pregnancy, different antibiotics have specific effect on changing maternal microbiota [19,20]. The latter could be transmitted to the fetus not only during delivery [21], but during pregnancy as well [21–23]. The placental microbiome produces metabolic and immune factors, and generates a balanced mucosal immune system [24]. When microbial exposure of fetus was disturbed, the risk of atopic immune responses and asthma is increased [25]. Thus, we speculate that the effect on asthma risk may be related to the antibacterial spectrum of penicillins or chloramphenicols, both of which target gram positive bacteria. Detailed types of microbiota and their explicit mechanism need to be further elucidated in future epidemiologic studies and in vitro experiments.

Our study also found that the early pregnancy may be a sensitive window for the exposure to penicillins or chloramphenicols. The early pregnancy is an important time for fetal immune development [26]. Even at the 12 weeks of gestation, T cell has begun to develop [26]. Furthermore, the sensitive window may also depend on the type of antibiotics. More research is needed. Since in CPP data, women were asked about their health condition 4 weeks before LMP, including medication, we used 4-week timeframe and were unable to restrict our analysis to exposure periods closer to the time of actual conception.

We acknowledge that the diagnostic criteria for asthma may have changed since 1960s. The asthma diagnosis then was based on notion that asthma is “a disease of the respiratory passages characterized by dyspnea of an obstructive type which is predominantly expiratory, reversible at least partially, and of varying severity and duration” [27]. Even today, asthma diagnosis in young children is still made based largely on symptom patterns and on a careful clinical assessment of family history and physical findings according to the Global Initiative for Asthma works [28,29]. This is due to the fact that lung function measures in children may not be reliable, even in children of 7 years [30]. In our study, we included only “definite asthma” confirmed by pediatricians and excluded suspected asthma. The prevalence of asthma was 5.5% at that time, suggesting that misclassification of asthma may not be a serious issue in our study.

It may be questioned whether the association between maternal antibiotic use before and during pregnancy and childhood asthma was due to confounding by indication or genetics, i.e., women who used antibiotics tended to have an elevated risk of asthma-related diseases. This tendency may pass down to their children. Thus, it would be the genetic susceptibility to asthma rather than antibiotic use per se that has caused this association [18]. Unfortunately, CPP did not collect information on indication for maternal antibiotic use. However, only maternal urinary tract infections during pregnancy were found to increase the risk of asthma in children in previous studies [31], whereas penicillins or chloramphenicols weren’t indicated to use in urinary tract infections. Therefore, the association between maternal exposure to penicillins or chloramphenicols and childhood asthma may not be confounded by maternal infections that led to penicillins or chloramphenicols use. Moreover, our study shows that even after we adjusted for history of maternal asthma and allergy, exposure to certain drugs, namely penicillins and chloramphenicols, during early pregnancy was associated with child asthma.

In addition, over 11,000 children were lost to follow-up and, therefore, often did not have asthma diagnosis. But the prevalence of antibiotic exposure or asthma history of mothers was similar between those lost and followed-up. For example, the prevalence of antibiotic exposure was 26.4% vs 27.7% for mothers included vs mothers with loss-to-follow-up, respectively. Likewise, the prevalence of maternal asthma history was 4.7% vs 5.3%, respectively. These differences appeared small and were unlikely to bias the results substantially.

In conclusion, maternal exposure to penicillins or chloramphenicols during pregnancy is associated with childhood asthma by 7 years of age. Early pregnancy may be a sensitive window. Our findings not only confirmed previous reports that maternal exposure to antibiotics enhances the risk of childhood asthma, but also showed that this association may be antibiotic-specific.

Supporting Information

Acknowledgments

This study was supported by grants from the National Natural Science Foundation of China (81530086 and81273091), the Ministry of Science and Technology of China (2014DFG31460) and Shanghai Municipality Science and Technology Commission (14XD1403300).