Antibiotic Exposure in Early Life Increases Risk of Childhood Obesity: A Systematic Review and Meta-Analysis

A number of studies have previously assessed the impact of antibiotic exposure in early life on the risk of childhood obesity, but no systematic assessment is currently available. A systematic review and meta-analysis was performed to comprehensively and quantitatively elucidate the risk of childhood obesity caused by antibiotic exposure in early life. Literature search was performed in PubMed, Embase, and Web of Science. Random-effect meta-analysis was used to pool the statistical estimates. Fifteen cohort studies involving 445,880 participants were finally included, and all those studies were performed in developed countries. Antibiotic exposure in early life significantly increased risk of childhood overweight [relative risk (RR) = 1.23, 95% confidence interval (CI) 1.13–1.35, P < 0.001] and childhood obesity (RR = 1.21, 95% CI 1.13–1.30, P < 0.001). Antibiotic exposure in early life also significantly increased the z-score of childhood body mass index (mean difference: 0.07, 95% CI 0.05–0.09, P < 0.00001). Importantly, there was an obvious dose–response relationship between antibiotic exposure in early life and childhood adiposity, with a 7% increment in the risk of overweight (RR = 1.07, 95% CI 1.01–1.15, P = 0.03) and a 6% increment in the risk of obesity (RR = 1.06, 95% CI 1.02–1.09, P < 0.001) for each additional course of antibiotic exposure. In conclusion, antibiotic exposure in early life significantly increases risk of childhood obesity. Moreover, current analyses are mainly taken from developed countries, and therefore the impact of antibiotic exposure on risk of childhood obesity in vulnerable populations or developing countries still needs to be evaluated in future studies.

A number of studies have previously assessed the impact of antibiotic exposure in early life on the risk of childhood obesity, but no systematic assessment is currently available. A systematic review and meta-analysis was performed to comprehensively and quantitatively elucidate the risk of childhood obesity caused by antibiotic exposure in early life. Literature search was performed in PubMed, Embase, and Web of Science. Random-effect meta-analysis was used to pool the statistical estimates. Fifteen cohort studies involving 445,880 participants were finally included, and all those studies were performed in developed countries. Antibiotic exposure in early life significantly increased risk of childhood overweight [relative risk (RR) = 1.23, 95% confidence interval (CI) 1.13-1.35, P < 0.001] and childhood obesity (RR = 1.21, 95% CI 1.13-1.30, P < 0.001). Antibiotic exposure in early life also significantly increased the z-score of childhood body mass index (mean difference: 0.07, 95% CI 0.05-0.09, P < 0.00001). Importantly, there was an obvious dose-response relationship between antibiotic exposure in early life and childhood adiposity, with a 7% increment in the risk of overweight (RR = 1.07, 95% CI 1.01-1.15, P = 0.03) and a 6% increment in the risk of obesity (RR = 1.06, 95% CI 1.02-1.09, P < 0.001) for each additional course of antibiotic exposure. In conclusion, antibiotic exposure in early life significantly increases risk of childhood obesity. Moreover, current analyses are mainly taken from developed countries, and therefore the impact of antibiotic exposure on risk of childhood obesity in vulnerable populations or developing countries still needs to be evaluated in future studies.
Keywords: antibiotics, childhood obesity, meta-analysis, risk factor, overweight inTrODUcTiOn Obesity has increased significantly among children and adolescents over the past several decades, and childhood adiposity has become a major challenge to public health worldwide (1,2). In the United States, the prevalence of obesity in children and adolescents is about 17.0% and it has also increased substantially (3). Excess adiposity during childhood is likely to persist into adulthood and further increases the predisposition to cardiovascular diseases, cancer, and early mortality (4,5). Childhood obesity also increases risk of other diseases during childhood, such as hypertension and depression (6)(7)(8). In addition, some obese children have exhibited early signs of cardiovascular dysfunction (9). The pathogenesis of obesity is complex and may involve many factors, and studies that explore the causes of the increasing epidemic of childhood obesity and possible solutions to this issue are necessary (10).
Emerging evidence has suggested that gut microbiota is linked to obesity and other metabolic disorders (11,12). There are obvious changes in gut microbiota of patients with obesity, and these changes may precede the clinical manifestation of obesity (13). Antibiotics can result in obvious alterations in the gut microbiota and even gut dysbiosis, which may further result in obesity or other metabolic disorders (12,14,15). Considering that antibiotics are frequently prescribed in pediatric patients, a precise assessment of the relationship between antibiotics and childhood obesity is very important (16). There are a number of epidemiological studies assessing the impact of antibiotic exposure in early life on childhood adiposity risk (17)(18)(19)(20)(21)(22)(23)(24). Several studies reported that antibiotic exposure could increase the risk of childhood overweight or obesity (17,19,23,24), while the other studies reported that antibiotic exposure had no obvious role on childhood adiposity risk (18,21). Besides, no systematic review and meta-analysis is currently available that provides a definite evaluation of the relationship between antibiotic exposure and childhood adiposity. An improved understanding of the issue has important implications for public health, because there is an increasing epidemic of childhood obesity and prudent use of antibiotics in early life may be one way to reduce this epidemic. Therefore, a systematic review and meta-analysis of available studies was performed to assess the relationship between antibiotic exposure in early life and childhood adiposity.

literature search
A literature search in PubMed, Emabse, and Web of Science was carried out to find articles that examined the impact of antibiotic exposure on childhood adiposity risk. The following terms were used: (antibiotics OR antibiotic OR tetracycline OR doxycycline OR cephalosporin OR penicillin OR metronidazole OR fluoroquinolone OR sulfonamide OR macrolide) AND (adiposity OR obesity OR obese OR overweight OR body mass index) AND (child OR children OR childhood OR boy* OR girl* OR infancy OR infants). All databases were searched from their inception through September 6, 2016, and an updated literature search was performed on February 26, 2017. Language restriction was not applied. Bibliographies of eligible articles and relevant reviews were also screened.

selection criteria
Selection criteria for the meta-analysis were as following: (1) prospective or retrospective cohort studies; (2) the outcomes of interest were the association between antibiotic exposure and childhood adiposity, or the impact of antibiotic exposure on childhood body mass index (BMI) or weight; (3) the exposure time was early life including infancy and prenatal period; (4) reporting risk estimates on the associations, such as relative risks (RRs) with 95% confidence intervals (95% CIs), or the difference in the z-scores of childhood BMI or weight between the exposed group and the non-exposed group. Infancy is usually defined as the period from birth to 2 years of age. Early life includes infancy and prenatal period, and is thus from fetal life through 2 years of age in the present study. Studies containing overlapping data, case reports, or case series were all excluded.

Data extraction and Quality evaluation
Two colleagues extracted data independently, and discrepancy was resolved by discussion. The extracted data mainly included first author, country, study design, antibiotic exposure, participants, follow-up time, adjusted variables, and outcomes of interest. If one study reported data of multiple follow-ups, we used the data for the longest follow-up time. The authors of included studies were contacted if crucial information was not provided in the articles.
The primary outcome of interest was the association of antibiotic exposure in early life with childhood overweight or obesity. The secondary outcome of interest was the difference of the z-scores of childhood BMI or weight between the exposed group and the non-exposed group. The evaluation of study quality was performed by the Newcastle-Ottawa scale, and studies scoring seven or more points were identified as high-quality studies (25).

Data analysis
The risk estimates on the associations between antibiotic exposure and childhood adiposity and the differences of childhood body weight between the exposed group and the non-exposed group were pooled using meta-analysis. Heterogeneity was measured using Cochran-Q test and I 2 method (26,27). Random-effect model was utilized to pool data in order to reduce the impact of heterogeneity across studies (28). Subgroup analyses were carried out by gender, exposure time, and study design. The study by Li et al. (29) reported data from two different comparisons, and we firstly used the risk estimate from the comparison between infants who used antibiotics and those who did not, and then  performed a sensitivity analysis by using the risk estimate from the comparison between infants who used antibiotics and those who had infections but did not receive antibiotics. Publication bias was evaluated by Egger's test and funnel plot (30). For doseresponse meta-analysis, the correlated log RR estimates across different courses of antibiotic exposure were calculated using generalized least-square regression method, and RRs were further pooled using random-effect meta-analysis (31,32). STATA (version 12.0) and Review Manager (version 5.2) were used for statistical analyses. P < 0.05 indicated a statistically significant difference.

resUlTs study selection and characteristics
From the literature search, we found a total of 1,294 abstracts, but only 26 studies were considered as potentially eligible and assessed by reading full-texts (17-24, 29, 33-49) (Figure 1). After detailed evaluation, 11 studies were further removed including 6 studies for irrelevant outcomes (34-36, 40, 42, 47), 3 studies for no usable data (38,43,44), and 2 studies for crosssectional design (37,46). The study by Korpela et al. only reported the r-value from linear correlation analysis, but did not report the difference in the z-score of childhood BMI or weight (43). Ultimately, 15 cohort studies involving 445,880 participants were included (17-24, 29, 33, 39, 41, 45, 48, 49) (Figure 1). Table 1 showed the characteristics of those 15 cohort studies ( Table 1). There were 12 prospective cohort studies, and 3 retrospective cohort studies ( Table 1). Those studies were published from 2011 to 2017. There was an obvious difference in the number of participants, which ranged from 97 to 260,556. All studies were performed in developed countries, and no study was from developing countries or was performed in predefined vulnerable populations (Table 1). Besides, about 40% of the studies were in the United States and represented over 75% of all subjects ( Table 1).

Meta-analysis
No obvious heterogeneity existed in those studies relating antibiotic exposure and childhood overweight (I 2 = 13.9%). Meta-analysis showed that antibiotic exposure in early life (from fetal life through age 2) significantly increased the risk of childhood overweight (RR = 1.23, 95% CI 1.13-1.35, P < 0.001) (Figure 2A). Subgroup analysis by exposure time suggested that both antibiotic exposure during infancy (from birth through age 2) and prenatal exposure to antibiotics could increase risk of childhood overweight ( Table 2). Other subgroup analyses by gender and time of antibiotic exposure during infancy also found increased risk of childhood overweight among those children with antibiotic exposure in early life ( Table 2).
There was moderate heterogeneity among studies relating antibiotic exposure and childhood obesity (I 2 = 46.8%). Metaanalysis showed that antibiotic exposure in early life (from fetal life through age 2) also independently increased childhood obesity risk (RR = 1.21, 95% CI 1.13-1.30, P < 0.001) (Figure 2B). When using the risk estimate of the comparison between infants who used antibiotics and those who had infections but did not  Figure 3). No heterogeneity existed in those studies relating antibiotic exposure and childhood BMI z-score (18,20,22,23,41,45,49) (I 2 = 7%). Meta-analysis showed that antibiotic exposure in early life could dramatically increase the z-score of childhood BMI (mean difference: 0.07, 95% CI 0.05-0.09, P < 0.00001) (Figure 2C). Antibiotic exposure could markedly increase the z-score of childhood BMI in both boys and girls (Table 2; Figure 3).
The funnel plot did not show an obvious risk of publication bias (Figure 4). PEgger' s test was 0.99 and it also proved the lack of publication bias. In addition, funnel plots in other meta-analyses on childhood obesity or childhood BMI also did not show an obvious risk of publication bias.

DiscUssiOn
The present study is the first systematic review comprehensively evaluating the impact of antibiotic exposure on childhood adiposity. Fifteen cohort studies involving 445,880 participants were finally included (17-24, 29, 33, 39, 41, 45, 48, 49). The results suggested that antibiotic exposure during early life significantly increased risk of childhood overweight and obesity, and there was an obvious dose-dependent relationship. In addition, antibiotic exposure also significantly increased the z-scores of childhood BMI and body weight. Thus, the meta-analysis provides strong evidence for the impact of antibiotic exposure during early life on childhood adiposity risk.
Another important issue is whether antibiotic exposure is an independent risk factor for childhood adiposity. In present metaanalysis, all included studies provided risk estimates adjusted for confounding factors ( Table 1). The pooled adjusted RRs for the associations of antibiotic exposure during early life with childhood overweight and obesity were both statistically significant ( Table 2). In addition, the z-scores of childhood BMI or body weight from included studies were also adjusted for confounding factors, and the pooled z-scores were statistically significant ( Table 2). Therefore, this meta-analysis proves that antibiotic exposure is an independent risk factor of childhood obesity.
The pathogenesis of obesity is complex and has not been well understood, and it is the same with childhood obesity (50,51). The role of gut microbiota in the pathogenesis of obesity is increasingly recognized, and accumulating evidence has suggested that gut microbiota has a causal role in obesity (50,52,53). Ussar et al. proposed that obesity was a result of interactions between gut microbiota, host genetics, and diet (54). However, the molecular mechanisms underlying the role of gut microbiota in obesity are not yet well known (55). More studies are still needed to further characterize the molecular mechanism of gut microbiota in modulating energy balance and obesity (56). Exploring the role of gut microbiota in childhood obesity  may also provide effective prevention or treatment strategies for childhood obesity (57). For example, a recent randomized controlled trial by Nicolucci et al. reported that probiotics can improve the eco-system imbalance of gut microbiotia and reduce the BMI z-score in children with overweight or obesity (58).
Data from animal studies have shown that antibiotic-induced changes in gut microbiota can result in fat accumulation by changing host metabolism (59). Antibiotics can also increase insulin resistance by changing gut microbiota (60,61). Cho et al. reported that antibiotics could alter the murine colonic microbiome and result in the development of obesity (62). Therefore, the important roles of gut microbiota in modulating energy balance and the alerted composition of gut microbiota caused by antibiotics provide some explanations for the association between antibiotic exposure and childhood adiposity. However, more researches are still needed to further explore the complex mechanism underlying this association. The increasing prevalence of childhood adiposity has become a major global health challenge, and actions to reduce the prevalence of obesity in children and adolescents are urgently needed (4,63,64). Investigating risk factors associated with childhood obesity can help us to develop effective prevention interventions (65). The meta-analysis suggests that antibiotic exposure can result in a substantially increased risk of childhood obesity. The finding has important implications in the clinical setting, since antibiotics are the most common drugs administered in infants. Clinicians should weigh the risk of subsequent childhood obesity associated with antibiotic exposure in early life when considering antibiotics in the absence of a clear indication.
The meta-analysis also indicates that prudent use of antibiotics may help to minimize the metabolic consequences of antibiotic exposure in early life. In addition, the approach that uses population analyses to understand pharmacokinetic/pharmacodynamic (PK/PD) parameters in children exposed to antibiotics is helpful to define optimal antibiotic dose and course duration, which undoubtedly can improve clinical outcomes and patient safety (66). To assess the impact of antibiotic exposure in early life on childhood obesity more adequately, population PK/PD analyses exploring the impact of antibiotics in different doses and course duration on body composition in children are recommended in future research.
Apart from the prudent use of antibiotics in infants, it is also important for us to develop some plausible methods to prevent obesity in those infants or children who have to receive the treatment of antibiotics. Several studies have studied some possible ways to prevent obesity in children who expose to antibiotics. Kaliannan et al. demonstrated a beneficial effect of omega-3 fatty acid in preventing antibiotic-induced gut dysbiosis and obesity (67). Economopoulos et al. found that co-administration of intestinal alkaline phosphatase with azithromycin prevented the development of metabolic diseases in mice, and the effect was mediated by alterations in gut microbiota (68). However, more studies are still needed to develop some effective methods to prevent the antibiotics-induced obesity in infants or children exposing to antibiotics.
The meta-analysis had several strengths. Firstly, the large number of participants could provide a precise evaluation of the relationship between antibiotics and childhood adiposity. In addition, the obvious dose-dependent relationship strengthened the evidence for the association. Finally, the consistency in the findings of several different types of outcomes further supported the robustness of the findings.
There were several limitations in the meta-analysis. Firstly, all of the current studies were taken from developed countries and there were missing data from developing countries or vulnerable populations. The current study is representative of developed countries, whereas the impact of antibiotics in children in vulnerable populations or developing countries may likely be different from that found in developed countries. Therefore, the impact between antibiotic exposure and childhood adiposity in developing countries or vulnerable populations is still unclear, and needs to be evaluated in future research. Secondly, there were obvious differences in those included studies. The confounding factors used in the adjusted estimates were different, which could result in heterogeneity and different outcomes (69). Thirdly, the use of antibiotics during follow-up may be an important factor modifying the impact of antibiotics on childhood adiposity risk. However, most included studies did not add antibiotics use during follow-up as a confounding factor, which might cause risk of bias. Fourthly, there were different types of antibiotics among included studies, and they might have different impact on childhood adiposity. Further studies are required to explore the impact of different types of antibiotics on childhood obesity. Finally, the relationship between antibiotic exposure and adiposity in adults has not been studied, and future research is recommended.
In summary, this meta-analysis provides strong evidence for the association between antibiotic exposure in early life and childhood adiposity. There is also an obvious dose-response relationship between antibiotic exposure and childhood adiposity. Prudent use of antibiotics is recommended for infants and children to reduce the risk of childhood adiposity. Future studies are also required to develop some effective methods to prevent obesity in infants or children exposing to antibiotics.

aUThOr cOnTribUTiOns
XS and JZ designed the study. XS, BW, XD, RS, and JZ contributed to the literature search, interpretation, writing, and proofreading of the manuscript. XS, BW, LL, and QY extracted data and performed data analyses. XS, BW, and XA revised the study. XS and BW generated the figures.

FUnDing
The present work was supported by grants from the National Natural Science Foundation of China (No. 81670722 and 81471004).