The antibiotic trimethoprim is an inhibitor of dihydrofolate reductase (DHFR), which converts folate to more active components [
We therefore tested the hypothesis that trimethoprim use during the months prior to pregnancy increases the risk of congenital malformations. To this end, we performed a nationwide study in which administrative registers of trimethoprim use were coupled with registrations of birth and congenital malformations.
Using the Danish Fertility Database, [
The Danish Fertility Database consists of individual-level data about the mother and the father, including a unique identification number, age, previous births, and abortions, as well as birth weight and length, death and cause of death, sex, and gestational age of the children [
Primary outcome was predefined as all congenital malformations (ICD-10 Q0.00–Q99.9). The secondary outcome was the subgroup of malformations. All major and minor malformations and malformation groupings are those of the European Surveillance of Congenital Anomalies (EUROCAT) classification system guide 1.3 [
Several different analyses were done. Offspring of mothers with no dispensed trimethoprim prescriptions in the 12 weeks before conception were used as the primary reference group as well as offspring of mothers with no dispensed prescriptions in the 12 weeks before pregnancy and during pregnancy. To avoid confounding by indication, we used the offspring of mothers with one or more dispensed prescriptions of the mecillinam, a penicillin used to treat urinary tract infections, during the 12 weeks before pregnancy as a reference group. Furthermore, we compared the results with the number of children diagnosed with a congenital malformation in women exposed to trimethoprim solely in weeks 13–24 and in weeks 25–36 before conception.
Comorbidity of mothers was assessed through the National Hospital Register on the basis of diagnoses made between 1995 and 2005. This was done to assess whether women treated with trimethoprim were more likely to be diagnosed with a teratogenic infection or conditions potentially harmful to the fetus. Diagnoses of infection, alcoholism, amniocentesis, and folate deficiency were included if they were given during pregnancy, whereas chronic conditions were included if the diagnoses were given before or during pregnancy. All prescribed drugs dispensed before and during pregnancy were classified according to the US Food and Drug Administration’s
All data management and analyses were performed using SAS software, version 9.1 (SAS Institute Inc., Cary, NC, USA). Three logistic regression models were used on dichotomous outcomes to estimate the odds ratios of congenital malformation. Model one was unadjusted; model two was adjusted for maternal age (as a continuous variable) and parity (as a continuous variable). We adjusted model three for maternal age (as a continuous variable), parity (as a continuous variable), income (four categories), education (four categories), usage of X-classified drugs (dichotomous variable), calendar time (1997-1998, 1999-2000, 2001-2002, 2003-2004), teratogenic infections (dichotomous variable, if the mother had at least one diagnosis of either syphilis, toxoplasmosis, rubella viruses, varicella virus, parvovirus B-19, cytomegalovirus, herpes simplex virus 1 and 2, or Venezuelan equine encephalitis virus during pregnancy), and folate deficiency, myasthenia gravis, virilizing tumors, diabetes mellitus, alcoholism, amniocentesis, and Sjogren’s syndrome (all as dichotomous variables). These variables had less than 0.5% missing values except for education (2.5% in the trimethoprim exposed group and 2.6% in the unexposed group (
For all analyses, a two-sided value of
All data were linked in computers held by Statistics Denmark and were made available with encrypted personal information [
We identified 521,267 births during the study period. A total of 402 children were born to 395 women exposed to a drug containing trimethoprim (trimethoprim or sulfamethoxazole plus trimethoprim) during the 12 weeks before conception (Table
Study population characteristics.
Use of trimethoprim in the 12-week period prior to pregnancy | No use of trimethoprim in the 12-week period prior to pregnancy | |
---|---|---|
|
|
|
Age (year) | 28.5 | 29.7 |
Parity (mean) | 1.80 | 1.85 |
Household income—number (%) | ||
<$50,000 | 76 (19) | 80,899 (16) |
$50,000–$100,000 | 194 (48) | 241,907 (46) |
$100,000–$150,000 | 110 (27) | 150,163 (29) |
>$150,000 | 22 (5) | 48,240 (9) |
Education—number (%) | ||
Low | 168 (42) | 182,072 (35) |
Medium | 123 (31) | 165,132 (32) |
Long | 98 (24) | 149,324 (29) |
No information available | 10 (2) | 13,333 (3) |
Comorbidityb—number (%) | ||
Teratogenic infectious diseasesc | 1a (0.3) | 133 (0.0) |
Folate deficiency | 0 (0) | 39 (0.0) |
Myasthenia gravis | 0 (0) | 38 (0.0) |
Virilizing tumors | 0 (0) | 84 (0.0) |
Diabetes Mellitus | 4 (1) | 6258 (1) |
Alcoholism | 0 (0) | 111 (0.0) |
Amniocentesis | 1a (0.3) | 5381 (1.0) |
Sjogren’s syndrome | 0 (0) | 38 (0.0) |
HIV/AIDS | 0 (0) | 60 (0.0) |
|
0 (0) | 2 (0.0) |
Previous organ transplantation | 1a (0.0) | 49 (0.0) |
Use of folic acid—number (%) | 9 (2) | 4296 (1) |
aNone had offspring with a congenital malformation. bAll information on comorbidity were based on diagnoses from the National Hospital Register. cSyphilis, toxoplasmosis, rubella virus, varicella virus, parvovirus B-19, cytomegalovirus, herpes simplex viruses 1 and 2, and Venezuelan equine encephalitis virus.
Of the children born to an exposed mother, 25 (6.2%) were diagnosed with a major congenital malformation compared with 17,465 (3.4%) among those born to unexposed mothers. Among the exposed women, 40 (10.0%) children were born with a diagnosed congenital malformation (including minor malformations) compared with 27,173 (5.2%) born to unexposed mothers.
The odds ratio (OR) for the occurrence of a major congenital malformation after exposure to trimethoprim during the 12 weeks before pregnancy (model 1) was 1.91 (95% confidence interval 1.28–2.87) compared to unexposed women. Adjusting for age and parity (model 2), the OR was 1.91 (1.27–2.86). Furthermore, we included usage of X-classified drugs, teratogenic infections, folate deficiency, myasthenia gravis, virilizing tumors, diabetes mellitus, alcoholism, amniocentesis, Sjogren’s syndrome, education, calendar time, income, and use of folic acid in addition to age and parity (model 3) in the regression model without any influence on the result (
Risk of congenital malformations associated with trimethoprim use during the 12-week period prior to conception.
Reference group | All unexposed children born 1997–2004 | All unexposed children born 1997–2004 adjusteda | Children born 1997–2004 by mothers with one or more prescription of mecillinam prior to pregnancy | Children born 1997–2004 |
---|---|---|---|---|
|
|
|
|
|
Major congenital malformations or (CI 95%) | 1.91 (1.28–2.87) | 1.87 (1.25–2.80) | 1.78 (1.15–2.77) | 2.05 (1.37–3.08) |
All congenital malformations or (CI 95%) | 2.01 (1.45–2.78) | 1.96 (1.41–2.72) | 1.90 (1.33–2.72) | 2.18 (1.57–3.03) |
aAdjusted for age, parity, usage of X-classified drugs, teratogenic infections, folate deficiency, myasthenia gravis, diabetes mellitus, alcoholism, amniocentesis, Sjogren’s syndrome, virilizing tumors, education, and income.
Of the children with major malformations in the trimethoprim group, seven had a malformation of the heart (
Types of major congenital malformations observed.
Type of major malformation | Number of malformations | Odds ratio | |
---|---|---|---|
Exposed | Unexposed | (95% CI) | |
Congenital malformations of the nervous system | 2 (0.5%) | 686 (0.1%) | 3.79 (0.94–15.25) |
Neural tube defects | 1 (0.3%) | 234 (0.0%) | 5.55 (0.78–39.68) |
Congenital malformations of the eye | 0 (0%) | 586 (0.1%) | |
Congenital malformations of the ear, face, and neck | 0 (0%) | 274 (0.1%) | |
Congenital malformations of the heart | 7 (1.7%) | 3682 (0.7%) | 2.49 (1.18–5.26) |
Congenital malformations of the respiratory system | 0 (0%) | 469 (0.1%) | |
Orofacial clefts | 2 (0.5%) | 1036 (0.2%) | 2.51 (0.62–10.09) |
Congenital malformations of the digestive system | 2 (0.5%) | 993 (0.2%) | 2.62 (0.65–10.52) |
Abdominal wall defects | 0 (0%) | 145 (0.0%) | |
Congenital malformations of the external genital organs | 1 (0.3%) | 1442 (0.3%) | 0.82 (0.13–6.40) |
Congenital malformations of the internal urinary system | 1 (0.3%) | 1273 (0.2%) | 1.02 (0.14–7.25) |
Congenital malformations of the limbs | 9 (2.2%) | 5416 (1.0%) | 2.18 (1.13–4.23) |
Congenital malformations of the musculoskeletal system | 1 (0.3%) | 798 (0.2%) | 1.62 (0.22–11.60) |
Other malformations | 1 (0.3%) | 662 (0.1%) | 1.96 (0.28–13.98) |
Teratogenic syndromes with malformations | 0 (0%) | 38 (0.0%) | |
Genetic syndromes and microdeletions | 1 (0.3%) | 315 (0.1%) | 4.62 (0.58–29.44) |
Chromosomal abnormalities | 1 (0.3%) | 653 (0.1%) | 1.99 (0.28–14.17) |
| |||
All major congenital malformations | 25 (6.2%) | 17,465 (3.4%) | 1.91 (1.28–2.87) |
All minor congenital malformations | 18 (4.5%) | 11,600 (2.2%) | 2.06 (1.28–3.31) |
All congenital malformations | 40 (10.0%) | 27,173 (5.2%) | 2.01 (1.45–2.78) |
Major congenital malformations among children of mothers exposed to trimethoprim in the 12-week period prior to conception and among unexposed according to the EUROCAT classification system.
Incidence of types of major congenital malformation in children of women exposed to trimethoprim before conception as compared to children of women unexposed to trimethoprim before conception. Only types of malformations reduced by folic acid supplementation before or during pregnancy are included in the figure. Odds ratio (95% CI).
In Denmark, trimethoprim exists in packages for 7.5 defined daily doses (DDD), 25 DDD, and in combination with sulfamethoxazole as 7.5 DDD. The 7.5 DDD trimethoprim package is the most common, accounting for 77% of exposures. To confirm that the findings relate to preconception exposure, we performed a logistic regression analysis that only included women redeeming prescriptions of 7.5 DDD trimethoprim and only 14–84 days before conception; this ensured a minimum of one-week washout period before conception. In this analysis, the odds ratio was 1.81 (1.11–2.96) compared to unexposed women in the same period.
Despite the very low statistical power, we analysed the effect of the three treatment regimes. Women with a prescription of 7.5 DDD trimethoprim (
There was no difference in the risk estimate when adjusting for age of offspring at the time of diagnosis and no association between the year of conception and the number of children born with a major malformation (
A secondary reference group of women exposed to mecillinam in the 12 weeks before conception was used in a direct comparison with the women exposed to trimethoprim in the same period. For this calculation, there was still an increased risk of having a child with a congenital malformation when exposed to trimethoprim compared to being exposed to mecillinam in the 12 weeks before conception (
There was no difference in the incidence of major congenital malformations in children born to women dispensing trimethoprim solely in weeks 13–24 (
A post hoc analysis regarding the women exposed to trimethoprim within the 50 days before conception was made and the odds ratio was only minimally different compared to the 12-week period prior to conception (
In the present study, we found a doubling in the prevalence of major congenital malformations in offspring of women exposed to trimethoprim during the 12 weeks before conception. According to the EUROCAT subgrouping, there was an increased prevalence of heart and limb defects.
We considered several sources of possible bias. Infections may be teratogenic, and in Denmark, for younger women, trimethoprim is almost exclusively used to treat urinary tract infections. To rule out the possibility that these infections are the cause of malformations, we compared our results with those from women receiving mecillinam, in Denmark, a widely used penicillin primarily to treat urinary tract infections.
The result of this analysis (Table
Multivitamin supplementation containing folic acid used prior to and during early pregnancy has been associated with a reduced risk of having offspring with malformations, amongst others, of the limbs and heart [
The mechanism of trimethoprim as a folic acid antagonist could explain the association found, and the previous studies have shown that trimethoprim can reduce serum folate levels [
In the present study, the period of interest was predefined as being 12 weeks but, in truth, we do not know how long the vulnerable period prior to pregnancy is. One previously published study has indicated that treatment with trimethoprim may lower plasma folate for up to 50 days [
Dispensing day of trimethoprim in the 12-week period prior to conception for mothers of children with major and minor malformations. Only women who dispensed one prescription of trimethoprim during the period and had no prescriptions for 85–336 days prior to pregnancy and 90 days after conception are included in the figure.
If the association between trimethoprim and malformations was due to induced folate deficiency, we would expect to find fewer children born with malformations in the study population in the years after introduction of an official folic acid supplementation policy. Official recommendations regarding folic acid supplementations were introduced in Denmark in 1997. At that time, it was estimated that only 5% of the Danish women received the recommended daily dosage of folate [
Prescription data and malformation data have been studied before and found to be accurately recorded [
The strengths of this study are the large number of cases and its nationwide coverage, including all women giving birth in Denmark during the study period. This ensures very high completeness of the data independent of age, race, and social, educational, and economic statuses, which thereby minimizes the selection bias. Furthermore, the study only includes women who obtained and paid for the medication at the pharmacy.
The main limitation of the study is a potential existence of unaccounted confounding. Women using trimethoprim preconceptionally may differ in important characteristics from women who do not receive the treatment. Such characteristic could be causally related to the pregnancy outcome and confound partially or entirely the observed association. Another limitation is the lack of information concerning dosage and compliance. Even though we know about package and tablet size, we have no precise information about the dose prescribed. For this reasons we have restricted the analysis to treatment or no treatment. Furthermore, we do not know about the indication for treatment, for which reason we cannot completely rule out the possibility that the results are confounded by indication. To meet and thereby minimize these limitations, we have adjusted the analyses for teratogenic conditions, age, parity, and so forth and used an alternative reference group of mecillinam users. Low compliance would bias the study towards a lack of effect and the estimated risk may therefore be underestimated. If the treatment was taken later, then the effect might be overestimated because of possible postconception administration. This would have only a minor effect since trimethoprim is mainly used for acute infections in Denmark. Furthermore, it has previously been estimated that the majority of prescription drugs redeemed are taken [
In model 3 we tried to adjust the analyses for conditions potentially harmful to the fetus. These conditions include medical conditions such as some autoimmune diseases and diabetes mellitus, infections during pregnancy potential harmful to the fetus, and diagnoses of folate deficiency. This information was taken from the National Hospital Register and only included women actually diagnosed with the conditions. If women had any of the conditions but were not diagnosed at a hospital, the information was not available to us. This is a limitation and could potentially confound our adjusted result. Furthermore, we adjusted for use of folic acid. This information is based on redeemed prescriptions on folic acid and is very deficient since the majority of folic acid and multivitamin containing folic acid used in Denmark are sold over the counter.
Trimethoprim is used by millions of fertile women worldwide. The World Health Organization (WHO) recommends that adults with symptomatic HIV or AIDS should use life-long daily prophylaxis with co-trimoxazole (trimethoprim and sulfamethoxazole) [
The scale of the problem could be even larger in some countries. In the United Kingdom, the Medicines and Healthcare Products Regulatory Agency has been discussing whether trimethoprim should be reclassified for over-the-counter availability [
Viewed together, although we find the results of the present study biologically plausible, it is the first time this hypothesis has been tested. It is important to test the hypothesis in other studies. Furthermore, it is important to keep in mind that prophylaxis with trimethoprim-sulfamethoxazole in people living with HIV or AIDS in sub-Saharan Africa has been shown to reduce morbidity and mortality [
In conclusion, we have found an association of maternal exposure to trimethoprim during the 12 weeks before conception and a doubling of the rate of congenital malformations in the subsequent offspring. This calls for further investigation, both epidemiological and in vivo animal studies.
All authors declare that they have no conflict interests to declare.
J. T. Andersen, M. Petersen, C. T. Pedersen, and H. E. Poulsen conceived and planned the work that led to the paper. J. T. Andersen and H. E. Poulsen obtained funding and J. T. Andersen wrote the paper. All authors played an important role in data analyses and interpretation of the results and made substantive suggestions for critically revision for important intellectual content and approved the final version of the paper.
The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the paper.
The study is supported by grants from the Danish Agency for Science, Technology and Innovation (