Antibiotics in Pregnancy and Celiac Disease Risk in Offspring
In this prospective population-based cohort study we used data from the ABIS cohort (All Babies in Southeast Sweden) in order to examine the association between antibiotic exposure in pregnancy and CD in the offspring.
Between October 1997 and October 1999, all parents to babies born in southeast Sweden were invited to participate in the ABIS cohort. Of the 21,700 babies born during the study period, the parents of 17,055 children (78.6%) gave their informed consent to participate. In the maternity ward (at childbirth), parents to 16,285 children completed a questionnaire that included questions on antibiotics use in pregnancy, maternal education level, heath status and first-degree heredity for CD, type 1 diabetes mellitus and other autoimmune diseases.
Parents were asked to complete a structured study diary during the child's first year of life reporting infectious diseases and feeding practice, including duration of breastfeeding and age at gluten introduction. The diaries were completed prospectively at home and collected when the child was one year of age. In our main analysis we included 8729 individuals with data on use of systemic antibiotics during pregnancy and with complete diary data for duration of breastfeeding and time of gluten introduction (see flow chart, Figure 1).
(Enlarge Image)
Figure 1.
Flow chart study participants.
The parents of the study participants were slightly more often born in Sweden and with a higher level of education, as compared with the source population of southeast Sweden. Additional background data of the ABIS cohort have been described elsewhere.
Data on CD were collected through contact with all paediatric departments (n = 8) in the ABIS study area. The majority of children with CD were identified through a study published in 2004. In 2007–2008, we again contacted the same paediatric departments and asked them to report additional ABIS children with a biopsy-verified CD (villous atrophy) diagnosed until December 1st 2006. In addition to a biopsy suggestive of CD (Marsh grade III), children with CD were required either to have CD-specific antibody markers or CD-consistent symptoms that resolved after introduction of gluten-free diet. In the current study, date of CD diagnosis equals date of first positive small-intestinal biopsy. The ABIS population was not actively screened for CD and therefore the children with CD were investigated due to clinical manifestations of possible CD.
We collected questionnaire data on use of any systemic antibiotics during pregnancy. The mothers exposed to antibiotics were asked to specify the name on the type of antibiotics. Mothers who were uncertain of their antibiotic use in pregnancy (n = 27) or who had only used non-systemic antibiotics (n = 2) were excluded from the study.
We used Cox regression to estimate Hazard Ratios (HRs) and 95% confidence intervals (CIs) for the risk of childhood CD according to maternal antibiotic exposure in pregnancy.
In our main analysis (Model A, n = 8729) we used diary data to adjust for duration of breastfeeding (0–2; 3–4; 5–6; 7–8; 9–10 and ≥11 months) and age at gluten introduction (0–2; 3–4; 5–6; 7–8; 9–10 and 11–12 months of age) as potential confounders for CD development.
In a second analysis (Model B, n = 8698) we also adjusted for maternal education level (≤12 vs. >12 years) and number of any parent-reported infectious disease in the child's first year of life, categorized into the following months of age: 0–2; 3–4; 5–6; 7–8; 9–10 and 11–12 months of age. Education level has been associated with antibiotic use and may influence the risk of CD diagnosis. Also infectious load in early childhood may be associated with subsequent CD. Follow-up began at child's birth and ended at time of CD diagnosis or December 1st 2006 (end of follow-up). However, because CD only develop after gluten introduction, which often occurs during the first year of life, we also performed a Cox regression in which the time scale began at one year of age until CD diagnosis or December 1st 2006 (end of follow-up).
Mode of delivery and gestational age pertain a great impact on the infant gut colonization and have been associated with an increased risk of CD in offspring. In two subanalyses we therefore restricted our data to term deliveries and children born through vaginal delivery, respectively. We also performed stratified analyses by infant sex and heredity for CD or type 1 diabetes mellitus.
Statistical significance was defined as 95% CIs not including 1.0. We used SPSS version 20.0 (SPSS, Inc, Chicago, IL) for the statistical analyses.
As opposed to short duration of breastfeeding, no breastfeeding at all can have a different effect on gut colonization. In a post-hoc analysis we therefore chose to adjust for duration of breastfeeding splitting the first time category (0–2 months) into: 0–6 days and 7 days-2 months (keeping 3–4 months; 5–6 months etc.). In a second post-hoc analysis we estimated the risk of CD in offspring according to antibiotic exposure in pregnancy excluding children with a first-degree relative with autoimmune disease. According to data from the at-birth questionnaire we defined heredity for an autoimmune disease as presence of at least one first-degree relative with any of the following conditions: Goitre with hypothyroidism or hyperthyroidism, Grave `s disease, pernicious anaemia, systemic lupus erythematosus, Addison's disease, any diabetes mellitus, gestational diabetes, CD, inflammatory bowel disease or rheumatoid arthritis.
To increase the statistical power of the study, we performed a post-hoc analysis including all 14,942 ABIS children with available data on antibiotic exposure in pregnancy. In this post-hoc analysis we used multiple imputation to replace missing values for infant nutrition data. However, since the statistics software SPSS cannot handle survival analyses with imputed data (the command "selection variable: rule: imputation > =1" only runs in linear regression and logistic regression), we first examined the odds ratio (OR) for future CD in our original dataset adjusting for infant nutrition (according to the covariates in Model A) before using multiple imputation to adjust infant feeding.
In a final post-hoc analysis we adjusted for the children's use of antibiotics during their first year of life. Data on antibiotic use was collected through a questionnaire when the child was one year old. The number of antibiotic courses used during the child's first year of life was classified into the following categories: no use; 1–2; 3–5; ≥6 courses of antibiotics. Regrettably, we largely lacked data on type of antibiotic agent used by the children.
At a significance level of 0.05 we had an 80% power to detect a relative risk of 2.45 for CD in offspring to mothers treated with antibiotics during pregnancy.
This study was as part of the ABIS study approved by the Research Ethics Committees of Linköping University (Li 287–96) and Lund University (Lu 83–97). Mothers gave their written informed consent after careful written as well as oral information and information via videotape.
Methods
In this prospective population-based cohort study we used data from the ABIS cohort (All Babies in Southeast Sweden) in order to examine the association between antibiotic exposure in pregnancy and CD in the offspring.
Study Population
Between October 1997 and October 1999, all parents to babies born in southeast Sweden were invited to participate in the ABIS cohort. Of the 21,700 babies born during the study period, the parents of 17,055 children (78.6%) gave their informed consent to participate. In the maternity ward (at childbirth), parents to 16,285 children completed a questionnaire that included questions on antibiotics use in pregnancy, maternal education level, heath status and first-degree heredity for CD, type 1 diabetes mellitus and other autoimmune diseases.
Parents were asked to complete a structured study diary during the child's first year of life reporting infectious diseases and feeding practice, including duration of breastfeeding and age at gluten introduction. The diaries were completed prospectively at home and collected when the child was one year of age. In our main analysis we included 8729 individuals with data on use of systemic antibiotics during pregnancy and with complete diary data for duration of breastfeeding and time of gluten introduction (see flow chart, Figure 1).
(Enlarge Image)
Figure 1.
Flow chart study participants.
The parents of the study participants were slightly more often born in Sweden and with a higher level of education, as compared with the source population of southeast Sweden. Additional background data of the ABIS cohort have been described elsewhere.
Coeliac Disease (CD)
Data on CD were collected through contact with all paediatric departments (n = 8) in the ABIS study area. The majority of children with CD were identified through a study published in 2004. In 2007–2008, we again contacted the same paediatric departments and asked them to report additional ABIS children with a biopsy-verified CD (villous atrophy) diagnosed until December 1st 2006. In addition to a biopsy suggestive of CD (Marsh grade III), children with CD were required either to have CD-specific antibody markers or CD-consistent symptoms that resolved after introduction of gluten-free diet. In the current study, date of CD diagnosis equals date of first positive small-intestinal biopsy. The ABIS population was not actively screened for CD and therefore the children with CD were investigated due to clinical manifestations of possible CD.
Maternal Antibiotic Exposure in Pregnancy
We collected questionnaire data on use of any systemic antibiotics during pregnancy. The mothers exposed to antibiotics were asked to specify the name on the type of antibiotics. Mothers who were uncertain of their antibiotic use in pregnancy (n = 27) or who had only used non-systemic antibiotics (n = 2) were excluded from the study.
Statistical Analyses
We used Cox regression to estimate Hazard Ratios (HRs) and 95% confidence intervals (CIs) for the risk of childhood CD according to maternal antibiotic exposure in pregnancy.
In our main analysis (Model A, n = 8729) we used diary data to adjust for duration of breastfeeding (0–2; 3–4; 5–6; 7–8; 9–10 and ≥11 months) and age at gluten introduction (0–2; 3–4; 5–6; 7–8; 9–10 and 11–12 months of age) as potential confounders for CD development.
In a second analysis (Model B, n = 8698) we also adjusted for maternal education level (≤12 vs. >12 years) and number of any parent-reported infectious disease in the child's first year of life, categorized into the following months of age: 0–2; 3–4; 5–6; 7–8; 9–10 and 11–12 months of age. Education level has been associated with antibiotic use and may influence the risk of CD diagnosis. Also infectious load in early childhood may be associated with subsequent CD. Follow-up began at child's birth and ended at time of CD diagnosis or December 1st 2006 (end of follow-up). However, because CD only develop after gluten introduction, which often occurs during the first year of life, we also performed a Cox regression in which the time scale began at one year of age until CD diagnosis or December 1st 2006 (end of follow-up).
Mode of delivery and gestational age pertain a great impact on the infant gut colonization and have been associated with an increased risk of CD in offspring. In two subanalyses we therefore restricted our data to term deliveries and children born through vaginal delivery, respectively. We also performed stratified analyses by infant sex and heredity for CD or type 1 diabetes mellitus.
Statistical significance was defined as 95% CIs not including 1.0. We used SPSS version 20.0 (SPSS, Inc, Chicago, IL) for the statistical analyses.
Post-hoc Analyses
As opposed to short duration of breastfeeding, no breastfeeding at all can have a different effect on gut colonization. In a post-hoc analysis we therefore chose to adjust for duration of breastfeeding splitting the first time category (0–2 months) into: 0–6 days and 7 days-2 months (keeping 3–4 months; 5–6 months etc.). In a second post-hoc analysis we estimated the risk of CD in offspring according to antibiotic exposure in pregnancy excluding children with a first-degree relative with autoimmune disease. According to data from the at-birth questionnaire we defined heredity for an autoimmune disease as presence of at least one first-degree relative with any of the following conditions: Goitre with hypothyroidism or hyperthyroidism, Grave `s disease, pernicious anaemia, systemic lupus erythematosus, Addison's disease, any diabetes mellitus, gestational diabetes, CD, inflammatory bowel disease or rheumatoid arthritis.
To increase the statistical power of the study, we performed a post-hoc analysis including all 14,942 ABIS children with available data on antibiotic exposure in pregnancy. In this post-hoc analysis we used multiple imputation to replace missing values for infant nutrition data. However, since the statistics software SPSS cannot handle survival analyses with imputed data (the command "selection variable: rule: imputation > =1" only runs in linear regression and logistic regression), we first examined the odds ratio (OR) for future CD in our original dataset adjusting for infant nutrition (according to the covariates in Model A) before using multiple imputation to adjust infant feeding.
In a final post-hoc analysis we adjusted for the children's use of antibiotics during their first year of life. Data on antibiotic use was collected through a questionnaire when the child was one year old. The number of antibiotic courses used during the child's first year of life was classified into the following categories: no use; 1–2; 3–5; ≥6 courses of antibiotics. Regrettably, we largely lacked data on type of antibiotic agent used by the children.
Power Calculation
At a significance level of 0.05 we had an 80% power to detect a relative risk of 2.45 for CD in offspring to mothers treated with antibiotics during pregnancy.
Ethics
This study was as part of the ABIS study approved by the Research Ethics Committees of Linköping University (Li 287–96) and Lund University (Lu 83–97). Mothers gave their written informed consent after careful written as well as oral information and information via videotape.
SHARE
