Introduction

Clarithromycin is a macrolide antibiotic used to treat common infections including respiratory tract infections, skin infections and helicobacter pylori infections. Only limited data are available concerning the effect of clarithromycin on the human fetus when used in pregnancy.

Animal studies have shown that clarithromycin can induce fetal loss in rabbits and monkeys when used in very low dosages and in high dosages, respectively. [1] One observational study concerning pregnant women showed a doubling of the number of miscarriages in women exposed to clarithromycin in early pregnancy compared to a match control group. [2] There is limited knowledge concerning the risk of congenital malformations among women exposed to clarithromycin during pregnancy.

Based on the current knowledge clarithromycin is not recommended for use in pregnancy. However, since use of clarithromycin is used in very common conditions and only half of pregnancies are planned a substantial number of women risk exposure to clarithromycin in early pregnancy. [3].

We therefore conducted a nationwide cohort study testing the hypothesis that use of clarithromycin in the first trimester is associated with miscarriage. Furthermore, we investigated whether there is an association between use of clarithromycin in the first trimester and major congenital malformations.

Methods

We identified all registered pregnancies in Denmark with a conception between 1 January 1997 and 31 March 2007 (n = 934 480). We excluded 2976 pregnancies due to coding errors.

From the Danish Fertility Database all live births with a conception date in the study period were identified. Using the National Hospital Register we identified all registered cases of miscarriages and provoked abortions (O02, O03, O04, O05 or O06 according to the International Classification of Diseases 10^th Danish revision). Information on major malformations was obtained from the National Hospital Register. All major malformations and subgrouping are according to the European Surveillance of Congenital Anomalies (EUROCAT) classification system guide 1.3. [4]

The Danish Fertility Database consists of individual-level data on the mother and 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 offspring. The time of conception is based on ultrasound estimates or information of the date of last menstruation. More than 99.5% of births in Denmark since 1978 are registered in the Danish Fertility Database. [5], [6] The National Hospital Register contains information on all hospitalizations in the country, including admittance data and discharge diagnosis. [7], [8] It holds more than 99% of discharge records from all Danish hospitals. [9] Since 1997, information on gestational length has been added to diagnoses of provoked abortion and miscarriage.

Information on prescription medication use was collected from the National Prescription Register (the Register of Medicinal Product Statistics). [10], [11] Exposure was defined as redemption of a prescription of a drug containing clarithromycin (Anatomical Therapeutic Chemical Classification (ATC) J01FA09) for systemic use.

The register contains individual-level data on all prescribed drugs dispensed at all pharmacies in Denmark since 1995. Pharmacies are required by law to register prescriptions and this activity is coupled with reimbursement of expenses from the state, which ensures highly accurate prescription data. Completeness has previously been estimated to be 97.5%. [12] The register has no information on indication of treatment or over-the-counter drugs.

The infection treated with clarithromycin, could itself induce miscarriages. To test for this possible confounding by indication we compared the hazard of miscarriage for redeeming a prescription of clarithromycin with the hazard of miscarriage when redeeming a prescription of phenoxymethylpenicillin (ATC J01CE02) for systemic use or erythromycin (ATC J01FA01) for systemic use in the first trimester. Clarithromycin is furthermore used in treatment of helicobacter pylori infections in combination with proton-pump inhibitors (PPI) (ATC A02BC) and amoxicillin (ATC J01CA04). We therefore tested whether redemption of prescriptions on these drugs in the first trimester was associated with miscarriages. Furthermore we compared the risk of miscarriage of women exposed to clarithromycin directly against the risk of women exposed to the four above mentioned drugs.

To exclude the importance of other characteristics of women using clarithromycin during pregnancy, we compared the result with the hazard of dispensing clarithromycin in the 12 weeks before pregnancy.

Results

We included 931 504 pregnancies identified within the study period. 77 553 (8.3%) miscarriages, 148 114 (15.9%) induced abortions and 705 837 (75.8%) live births.

Women redeeming a prescription of clarithromycin in the study period were older (p = 0.003) and had a lower educational level (<0.0001) compared to unexposed women.

Major Congenital Malformations

Among the 401 women exposed to clarithromycin in the first trimester 253 resulted in live births. Among these, nine children were diagnosed with a major malformation (3.6%) compared to 24 808 (3.5%) among children born by unexposed mothers (table 1). We found an unadjusted odds ratio of 1.01 (CI95% 0.52–1.97) of having a child diagnosed with a major malformation when exposed to clarithromycin in the first trimester compared to unexposed women, and an adjusted odds ratio of 1.03 (CI95% 0.53–2.00). There was no significant difference in the number of major malformation according to the EUROCAT grouping system between women exposed to clarithromycin and unexposed women (table 2).

Download:

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

Table 1. Basic characteristics.

https://doi.org/10.1371/journal.pone.0053327.t001

Download:

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

Table 2. Exposure to clarithromycin in the first trimester.

https://doi.org/10.1371/journal.pone.0053327.t002

Other Analyses

To investigate if the increased hazard of having a miscarriage when exposed to clarithromycin could be confounded by indication, we tested for the hazard of some drugs typically used for respiratory infections in Denmark. We found no increased hazard of having a miscarriage after redeeming a prescription, in the first trimester, of erythromycin (HR = 1.00 (CI95% 0.92–1.10)) or phenoxymethylpenicillin (HR = 1.02 (CI95% 0.98–1.07)). Furthermore, we investigated if use of PPI or amoxicillin, both used as first line treatment in combination with clarithromycin in helicobacter pylori infection, were associated with miscarriages. We found no increased HR for PPI or amoxicillin (table 3, figure 1). Furthermore we estimated the hazard of miscarriage in women exposed to clarithromycin compared directly to women exposed to the drugs used in same conditions as clarithromycin. We found that women exposed to clarithromycin had an adjusted hazard of 1.51 (CI95% 1.09–2.12) of having a miscarriage compared directly to the hazard of miscarriage in women exposed to phenoxymethylpenicillin, and 1.45 (CI95% 1.04–2.03) compared directly to the hazard of miscarriage in women exposed to erythromycin (figure 2). Women exposed clarithromycin had an adjusted hazard of 1.69 (CI95% 1.19–2.39) of having a miscarriage compared directly to women exposed to PPIs and an adjusted hazard of 1.88 (CI95% 1.34–2.65) compared directly to women exposed to amoxicillin (figure 2).

Download:

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

Figure 1. The hazard ratio of having a miscarriage when redeeming a prescription of a drug in the first trimester of pregnancy.

All hazards are adjusted for age, parity, educational level, and income. *Proton pump inhibitors.

https://doi.org/10.1371/journal.pone.0053327.g001

Download:

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

Figure 2. The hazard ratio of having a miscarriage when redeeming a prescription of clarithromycin in the first trimester of pregnancy directly compared to the hazard of women redeeming prescriptions of different drugs used in similar conditions.

All hazards are adjusted for age, parity, educational level, and income. *Proton pump inhibitors.

https://doi.org/10.1371/journal.pone.0053327.g002

Download:

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

Table 3. Sensitivity analyses. Adjusted hazard ratios of having a miscarriage when redeeming a prescription in the first trimester compared to unexposed.

https://doi.org/10.1371/journal.pone.0053327.t003

There was no increased hazard of having a miscarriage when being exposed to clarithromycin in the 12 weeks before conception (HR = 0.90 (CI95% 0.77–1.06)).

Discussion

We found an increased hazard of having a miscarriage when redeeming a prescription of clarithromycin in the first trimester of pregnancy. We did not find an association between exposure to clarithromycin and major congenital malformations.

The findings of this study strengthen other available data. In an animal study an increased number of fetal losses among rabbits given an intravenous dose 17 times lower than the maximum proposed human dose. [1] In monkeys, clarithromycin plasma levels three times the human serum level, resulted in embryonic loss. [1] An observational study showed that women treated with clarithromycin in the first trimester had a doubling of the number of miscarriages compared to a control group. Although the finding was statistical significant compared to the control group the authors noted that the number of miscarriages in the exposed group was within the expected range and they conclude that clarithromycin can be taken safely in pregnancy. [2]

To assess the robustness of our findings we analysed the hazard of having a miscarriage when redeeming a prescription of phenoxymethylpenicillin and erythromycin. These are two antibiotics commonly used in Denmark to treat respiratory infections. We did not find increased hazard ratios of miscarriages in these groups indicating that infections requiring these treatments are not associated with miscarriages nor do the drugs induce miscarriages themselves. It has previously been shown that the transplacental transfer of clarithromycin is two times that of erythromycin. [14], [15] This could potentially explain why we find an increased hazard for clarithromycin but not for erythromycin. Furthermore, we analyzed the hazard of having a miscarriage when redeeming a prescription of PPIs or amoxicillin. Both drugs are used in the treatment of helicobacter pylori infections in combination with clarithromycin. We did not find an increased hazard of miscarriage when redeeming prescriptions on these drugs, which indicates that PPIs and amoxicillin and infections treated with these drugs are not, associated with miscarriages. The hazards of phenoxymethylpenicillin, erythromycin, PPIs and amoxicillin suggest that the primary result regarding clarithromycin and miscarriages are not a product of confounding by indication.

To address the importance of other characteristics of women using clarithromycin, we compared the result with dispensing of clarithromycin much earlier than the pregnancy period but found no increased prevalence with dispensing in the 12 weeks before pregnancy. This indicates that the increased hazard of having a miscarriage is not explained by unaccounted characteristics of exposed women.

We did not find an association between first trimester exposure to clarithromycin and major congenital malformations. All though this is a small study with limited power, the result is supported by three other studies. [2], [16], [17] A higher rate of heart defects has been observed among rat offspring when the mother was given clarithromycin during pregnancy compared to unexposed controls. [1] Likewise, studies on mice demonstrated a higher rate of offspring with cleft palate when given clarithromycin compared to controls. [1] None of the clarithromycin exposed in our study were given a diagnosis of oro-facial cleft and we did not find an increased odds ratio of being diagnosed with a heart defect when exposed to clarithromycin in the first trimester compared to unexposed.

The main limitation of the present study is the lack of information concerning treatment indication and we can therefore not completely rule out that our results are confounded by indications. Furthermore, we do not have information on the prescribed dosage and therefore the analyses are restricted to treatment or no treatment. Even though the women went to the pharmacy, redeemed the prescription and paid for the drug we do not know whether or not they were exposed. If they were not exposed, our results would be biased towards underestimating the effect. Furthermore there is a risk of bias due to unaccounted confounders; even though we analysed the risk miscarriage of women exposed to clarithromycin before pregnancy we cannot completely rule out that women exposed to clarithromycin might differ from unexposed women in aspects causally related to the outcome. This could include obesity, alcohol consumption, smoking and antiphospholipid syndrome.

Another limitation is the low number of women exposed to clarithromycin in the study period. Analyses of risk of different subgroups of major congenital malformations should therefore be interpreted with caution.

We only have information on the date the miscarriage was recognized by the women and unfortunately not the exact time when the fetus perished in the uterus. If the exposure to clarithromycin was before the actual miscarriage, but after the fetus perished, the analyses could be biased.

We used registers to define the outcomes and they may be subject to misclassification. If women experience a miscarriage without contacting a hospital the number of registered miscarriages would be underestimated. This underreporting has been estimated to 30% [18] and is probably due to miscarriages early in pregnancy. We find no indications of a difference in the reporting of miscarriages between the two groups. If the increased hazard of miscarriages in women treated for an infection during the first trimester should be due to better reporting we would expect to find the same increase in the women treated with erythromycin, penicillin and amoxicillin, which we did not (table 3). The quality of the malformation diagnoses has been validated and found to have a predictive value of 88% for having a congenital malformation, with a completeness of 90% and misclassification, if any, was most probably random. [19].

This study covers nationwide data and includes all registered birth, provoked abortions and miscarriages in the study period. We thereby minimize selection bias, and ensure high completeness of data independent of social-, economic-, and educational level, age and race.

The registers have previously been validated and found to be accurately recorded. More than 99.5% of all birth in Denmark have been recorded in the Danish Fertility Database [5], [6] and more than 99% of all discharge diagnoses have been registered in the National Hospital Register. [20].

Exposure is based on information from the National Prescription Register which holds records on all redeemed prescriptions which means that the drugs have been collected at the pharmacy and paid for. Completeness has been estimated to 98%. [12].

This study demonstrates that treatment with clarithromycin in the first trimester of pregnancy was associated with a 50% increase (hazard ratio) of miscarriage compared to no treatment and a similarly increased risk compared to penicillin and erythromycin. We did not find an association between clarithromycin and major congenital malformations. The results are similar to other studies addressing the same issue both in human and animals. However, further research is required to explore the possible effect of treatment indication on the associations found.