Safety of oseltamivir during pregnancy: a comparative study using the EFEMERIS database

Authors


Abstract

Objective

To compare pregnancy outcome between women exposed and unexposed to oseltamivir during pregnancy.

Design

A comparative observational cohort study of women exposed to oseltamivir during pregnancy.

Setting

A French prescription database (EFEMERIS) that includes data for pregnant women was used. EFEMERIS records prescribed and dispensed reimbursed drugs during pregnancy and pregnancy outcomes in Haute-Garonne, South West France.

Population

Women who delivered from 1 July 2004 to 31 December 2010.

Methods

The study compared exposed and unexposed pregnant women. Two women unexposed to oseltamivir were individually matched, by maternal age, month, and year of delivery, with one women exposed to oseltamivir. Multivariable conditional logistic regression and multivariable Cox proportional hazards regression were used to evaluate associations between each outcome and exposure to oseltamivir during pregnancy.

Main outcome measures

Pregnancy loss for any cause, preterm delivery, low birthweight, neonatal pathology, and congenital malformation.

Results

A cohort of 337 (0.58% of women included in EFEMERIS) women exposed to oseltamivir were compared with 674 unexposed women. The risk for pregnancy loss (HR 1.52; 95 % CI 0.80–2.91), for preterm birth (adjusted OR 0.64; 95% CI 0.31–1.27), and for neonatal pathology (adjusted OR 0.62; 95% CI 0.23–1.54) did not differ between exposed and unexposed groups. When exposure during organogenesis was considered, one case of congenital anomaly (2.0%) among 49 exposed women and one case (1.0%) among 99 unexposed women were observed (crude OR 2.00; 95% CI 0.13–32.00).

Conclusions

There was no significant association between adverse pregnancy outcomes and exposure to oseltamivir during pregnancy.

Introduction

Pregnant women are at increased risk for severe illness, hospitalisation, and complications as a result of influenza, especially during the second and third trimesters. An increased risk for miscarriage, stillbirth, and preterm birth has also been reported in pregnant women with influenza infection.[1-5]

Oseltamivir (Tamiflu; F. Hoffman-La Roche Ltd., Basel, Switzerland) and zanamivir (Relenza; GlaxoSmithKline Inc., Ontario, Canada) are neuraminidase inhibitors recommended for the treatment and prophylaxis of influenza types A and B.[6, 7] They have been used to treat influenza since 2006, and their use has increased with the spread of the influenza A (H1N1) pandemic in 2009. Data suggest a moderate benefit: treatment with antivirals, if taken within 48 hours of the onset of symptoms, reduced the duration of the illness by ~1 day only.[7-9] Even if there are limited safety data, it was assumed that the moderate maternal benefit justified their use during pregnancy. Current French health authority guidelines recommend the use of neuraminidase inhibitors for the treatment and prophylaxis of influenza in women at any time of pregnancy.

Recently, a few studies have investigated the potential adverse effects of the use of neuraminidase inhibitors during pregnancy on the fetus.[10-14] These studies found no increased risk for small for gestational age (SGA), low birthweight, preterm delivery, congenital malformation, and stillbirth; however, considering the relatively low number of women exposed to antiviral drugs, more comparative studies are needed to confirm the safety of their use during pregnancy.

This study was designed to evaluate the potential effects of oseltamivir used in prophylaxis and treatment of influenza in women on pregnancy outcomes.

Methods

This observational cohort study compared two groups of pregnant women: one group exposed to oseltamivir during pregnancy and one group unexposed to this drug. The study was based on EFEMERIS, a French prescription database that includes records for pregnant women.[15]

EFEMERIS is a French database holding data on prescriptions in the general population, and can be used to evaluate drug risks during pregnancy. Data come from three different sources. Firstly, the French Health Insurance System (Caisse Primaire d'Assurance Maladie, CPAM) from Haute-Garonne (in South West France) records the reimbursed drugs prescribed and dispensed to patients under general state coverage (80% of the Haute-Garonne population). Secondly, the Mother and Child Protection Centre (Protection Maternelle et Infantile, PMI) records data from the children's certificates that are filled out during compulsory medical examinations at 8 days, 9 months, and 2 years. These health certificates contain data for both the mother (maternal characteristics, some pathologies during pregnancy) and the child (child characteristics, neonatal pathology, congenital malformation). Thirdly, the Antenatal Diagnosis Centre (Centre de Diagnostic Anténatal, CDA) centralises all occurrences of major and minor malformations in the maternities of the region if a therapeutic termination has been considered. The mother and her outcome (newborn or pregnancy loss) are linked with an anonymous irreversible code. More details on the design of EFEMERIS were published previously.[15] All drugs are classified according to the World Health Organization's Anatomical Therapeutic Chemical classification. The EuroCAT (European Network for Surveillance of Congenital Anomalies) classification system is used to classify all congenital malformations (major and minor). The date of conception is transmitted by the CPAM (all expecting mothers have to declare their pregnancy). The date of birth is given by the CPAM and the CDA. The timing of exposure was calculated with the date of conception and the date of medicine dispensation.

At the time of the study, 58,034 mother–outcome pairs with women who delivered in Haute-Garonne between 1 July 2004 and 31 December 2010 were recorded in EFEMERIS. The study was approved by the French Data protection Agency (Commission Nationale des Informations et Libertés, CNIL).

Any mother–outcome pair who received at least one prescription of oseltamivir during pregnancy were assigned to the ‘exposed’ group. The single woman receiving a prescription of zanamivir was not included in the study. Two ‘unexposed’ women for each ‘exposed’ woman were randomly selected from the individuals in the study population who did not receive a prescription of oseltamivir during pregnancy. ‘Unexposed’ women were individually matched with the ‘exposed’ women by maternal age, and month and year of the end of the pregnancy.

Maternal characteristics considered as potential confounding factors were multiple births, the presence of a long-term adverse health condition, pre-eclampsia, gestational diabetes, pregnancy-induced hypertension, a number of different active substances prescribed during pregnancy, the use of folic acid during organogenesis, the use of teratogens (retinoids, antiepileptics, vitamin K antagonists) during organogenesis (exposure between days 1 and 56 of gestation, which is the period of high susceptibility to teratogens).

We examined the relationship between oseltamivir exposure during pregnancy and pregnancy outcomes, including pregnancy loss for any cause (legal or therapeutic termination, miscarriage, stillbirth, and ectopic pregnancy), the occurrence of congenital malformation (major and minor congenital malformations could be identified on the 8-day and 9-month child health certificates, or from the CDA in the case of a therapeutic termination), preterm birth (<37 completed weeks of gestation), low birthweight (<2500 g), and the occurrence of neonatal pathology (identified through the 8-day child health certificates, for example respiratory distress, pneumothorax, neonatal jaundice, metabolic disorders, or sepsis). The analysis of risk for preterm delivery was restricted to exposure before 37 weeks of gestation, the analysis of risk for congenital malformation focused on exposure during organogenesis, and the analysis of risk for neonatal pathology also focused on exposure during the third trimester of pregnancy.

Continuous variables are expressed as means (± SDs), and categorical variables are expressed as percentages. A comparison of quantitative variables or proportions was performed among the two groups (exposed and unexposed) using simple conditional logistic regression. When exposure was restricted to a smaller period than the whole pregnancy, percentages were only calculated from the completed matched pairs. We used conditional logistic regression to analyse the effect of exposure to oseltamivir on the risks for each outcome by calculating unadjusted odds ratios (ORs), along with 95% confidence intervals (95% CIs). To identify confounding factors, we used simple conditional logistic regression to evaluate associations between each potential confounding factor and each outcome. Only variables with P < 0.2 and that were clinically significant were selected for inclusion in the multivariate model. We subsequently used conditional forward stepwise multivariate logistic regression to calculate adjusted odds ratios (aORs), along with 95% CIs. For rare outcomes, exact conditional logistic regression was calculated. Moreover, in order to take into account the time-dependent characteristic of exposure, we used survival analysis with time-dependent variable.[16] Cox proportional hazards regression was used to estimate hazard ratios with 95% CIs, comparing the hazard rates of pregnancy loss between exposed and unexposed women. Exposure was treated as a time-dependent variable. A woman was assigned to the ‘exposed’ group from the date of the first prescription for oseltamivir to the end of the pregnancy, but was assigned to the ‘unexposed’ group prior to the first prescription or if she did not receive any prescription for oseltamivir during pregnancy. Gestational days were used as the underlying time.

With a sample size of 300 women in the exposed group and 600 women in the unexposed group, and assuming a 5–10% outcome prevalence among the unexposed group, our data provided 80% power to detect an effect with OR = 2. All statistical tests were two-tailed, with P < 0.05 considered to represent statistical significance. Missing data were excluded from the analysis. All analyses were performed using sas 9.2 (SAS Institute, Cary, NC, USA).

Results

In the EFEMERIS database, a total of 337 of 58,034 (0.58%) mothers received at least one prescription of oseltamivir during pregnancy. This represented 348 prescriptions, most of which were dispensed in 2009 (98%): 37% of the exposures were in the second trimester, 32% were in the first trimester, and 31% were in the third trimester. Sixteen per cent of prescriptions were dispensed during organogenesis (between days 1 and 56 of gestation). The 337 exposed women were compared with 674 unexposed pregnant women.

Table 1 shows some maternal characteristics, morbidities, and drug use, comparing antiviral-exposed and unexposed women. The average maternal age at delivery was 30.2 years (SD 5.4 years). A significantly higher number of active substances prescribed during pregnancy was observed in the exposed group compared with the unexposed group (12.5 ± 7.0 versus 9.7 ± 6.9, P < .0001). Using the Anatomical Therapeutic Chemical (ATC) classification system, the study emphasised that compared with the unexposed group, antiviral-exposed women received significantly more prescriptions from the following drug classes: alimentary tract and metabolism (89.9 versus 78.2%); nervous system (86.7 versus 68.4%); respiratory system (67.7 versus 50.2%); blood and blood-forming organs (63.8 versus 56.8%); anti-infectives for systemic use (56.4 versus 47.0%); dermatologicals (50.7 versus 43.6%); genito-urinary system and sex hormones (35.3 versus 28.3%); systemic hormonal preparations (23.7 versus 15.1%); and musculoskeletal system (16.9 versus 12.5%).

Table 1. Maternal characteristics, comorbidities, and drug use among women exposed and unexposed to oseltamivir during pregnancy in the EFEMERIS cohort (France, 2004–2010)
CharacteristicsAntiviral-exposed (n = 337)Unexposeda (n = 674) P
  1. Data are represented as n (%), mean ± standard deviation.

  2. a

    Matched with exposed group according to maternal age, and month and year of delivery.

  3. b

    Simple conditional logistic regression.

  4. c

    Simple exact conditional logistic regression.

  5. d

    Received during pregnancy, oseltamivir excluded.

  6. e

    Live births only.

Multiple pregnancies, n 327633 
Multiple pregnancies b 8 (2.4%)19 (3.0%)0.71
Long-term adverse health condition, n 337674 
Long-term adverse health condition b 9 (2.7%)19 (2.8%)0.89
Pre-eclampsia, n 320602 
Pre-eclampsia c 2 (0.6%)2 (0.3%)0.81
Gestational diabetes, n 320602 
Gestational diabetes b 8 (2.5%)22 (3.7%)0.74
Hypertension, n 320602 
Hypertension b 2 (0.6%)2 (0.3%)0.81
Drugs use, n 337674 
Number of different active substances prescribedb, d12.5 ± 7.09.7 ± 6.9<0.0001
Organogenesis folic acid useb81 (24.0%)148 (22.0%)0.45
Organogenesis antiepileptic usec1 (0.3%)4 (0.6%)0.54
Organogenesis retinoid usec5 (1.5%)0 (0.0%)
Teratogen substances used during organogenesis (retinoids, antiepileptics, vitamin K antagonists)c6 (1.8%)4 (0.6%)0.09
Gestational age (weeks), n e 326631 
Gestational age (weeks) b 39.5 ± 1.739.5 ± 1.80.96

There were fewer pregnancy losses, for any cause, among the women exposed to oseltamivir during pregnancy compared with the unexposed group (3.6 versus 6.7%; Table 2). After adjusting for duration of exposure using Cox proportional hazards regression, no significant difference between women exposed and unexposed to oseltamivir during pregnancy was observed (HR 1.52; 95% CI 0.80–2.91; Table 3).

Table 2. Fetal and neonatal outcomes by exposure status in EFEMERIS cohort (France, 2004–2010)
OutcomesAntiviral-exposed women (n = 337)Unexposed womena (n = 674)
  1. a

    Matched with exposed group by maternal age, and month and year of delivery.

Pregnancy loss
Yes12 (3.6%)45 (6.7%)
No325 (96.4%)629 (93.3%)
Missing00
Preterm delivery
Yes18 (5.8%)45 (7.5%)
No292 (94.2%)554 (92.5%)
Missing2773
Low birthweight
Yes16 (5.1%)35 (5.9%)
No299 (94.9%)555 (94.1%)
Missing2284
Neonatal pathology
Yes9 (2.8%)29 (4.8%)
No311 (97.2%)573 (95.2%)
Missing1772
Neonatal pathology (third trimester exposure)
Yes1 (0.9%)10 (5.2%)
No105 (99.1%)183 (94.8%)
Missing019
Congenital malformation (organogenesis exposure)
Yes1 (2.0%)1 (1.0%)
No48 (98.0%)98 (99.0%)
Missing35
Table 3. Fetal and neonatal outcomes: crude and adjusted odds ratios according to oseltamivir exposure during pregnancy in the EFEMERIS cohort (France, 2004–2010)
OutcomesMeasure of Association (95% CI)a
CrudeAdjusted
  1. a

    For pregnancy loss, the reported measures of association are hazard ratios; for all other outcomes, the measures of association are odds ratios.

  2. b

    Adjusted for multiple births.

  3. c

    Adjusted for preterm births.

Pregnancy loss1.52 (0.80–2.91)
Preterm deliveryb0.75 (0.42–1.33)0.64 (0.31–1.27)
Low birth weightc0.81 (0.43–1.54)0.38 (0.07–1.39)
Neonatal pathologyc0.57 (0.27–1.24)0.62 (0.23–1.54)
Neonatal pathology (third-trimester exposure)c0.17 (0.02–1.25)0.30 (0.04–2.50)
Congenital malformation (organogenesis exposure)2.00 (0.13–32.00)

No difference in the rates of preterm delivery and low birthweight between the exposed group and the unexposed group was observed (Tables 2 and 3). There was no significant association between neonatal pathology with exposure to oseltamivir during pregnancy (aOR 0.62; 95% CI 0.23–1.54; Table 3). The study led to the same conclusion when only exposure during the third trimester was considered (Table 3).

If only exposure during organogenesis was considered, just one case of the 49 (2.0%) among the exposed group and just one case of the 99 (1.0%) among the unexposed group were observed (crude OR 2.00; 95% CI 0.13–32.00; Tables 2 and 3). The malformation in the exposed infant was a congenital heart defect. The congenital heart defect was noted in the 2-year child health certificate of a male baby. The 26-year-old mother received oseltamivir during the first month of pregnancy. The mother took several other drugs during pregnancy: salicylic acid, paracetamol, tuaminoheptane, escitalopram, and oxomemazine during the first trimester; paracetamol, amoxicillin, budesonide, tixocortol, and some homeopathies during the third trimester; and none during the second trimester.

Discussion

Main findings

Our study did not find any significant association between exposure to oseltamivir during pregnancy and pregnancy loss for any cause, preterm delivery, low birthweight, neonatal pathology, and congenital malformation.

Strengths and weaknesses

The present study is one of the largest comparative studies of adverse fetal outcomes, with 337 women exposed to oseltamivir during pregnancy. The major methodological strengths of EFEMERIS database include the fact that data are linked at an individual level. Information about the outcomes and medications are prospectively registered, thereby excluding recall bias. In France oseltamivir can only be dispensed with a prescription, so we have an almost complete coverage of the exposure. Moreover, data contain real dates of conception and pregnancy end. Furthermore, ~70% of the study infants are followed for up to 9 months, which facilitates the late detection of congenital malformations.

By contrast, our study suffers from some methodological insufficiencies inherent in a pharmacoepidemiological study. First, we cannot guarantee that all of the women who had a prescription for oseltamivir actually consumed their medicine. Furthermore, our study did not detect women who received all of their treatment from a hospital pharmacy during hospitalisation; however, when treatment was to be continued after hospitalisation, women were identified by the medicines delivered in outpatient pharmacies. Moreover, as a prescription database EFEMERIS does not provide any hospital diagnoses. So it is not possible to separate the possible effects of the flu itself from the effects of the medicine, as we do not know why a woman has been treated. Furthermore, some other unmeasured confounding factors may have influenced the results, despite our multivariate analysis. For example, potential confounding factors such as medical and obstetric history and pre-existing maternal pathologies could not have been evaluated in this study. Data on the mother's smoking habits and alcohol consumption are available from the child health certificates but, as is often the case in such studies, these are not very reliable. In order to evaluate the risk for pregnancy loss for any cause in the exposed group compared with the unexposed group, we used a Cox model with time-dependent exposure to avoid an immortal time bias among the exposed women. For such a model, no significant association between oseltamivir exposure during pregnancy and pregnancy loss for any cause was highlighted. However, although data regarding the risks are reassuring, the data set would have only been able to detect a 1.9-fold increased risk for preterm birth, a 2.0-fold increased risk for low birthweight, a 2.1-fold increased risk for neonatal pathology, and a 3.0-fold increased risk for congenital malformation in the exposed women. Thus, it would be of interest to perform further studies including a larger number of exposed women by using, for example, international multicentre analysis.

Interpretation

Reproductive studies with rats and rabbits did not demonstrate any teratogen or fetotoxic effect of oseltamivir or zanamivir; however, a dose-dependent increase in the incidence rates of skeletal malformation was notified with high fetal exposure to oseltamivir.[17] In our study, only one skeletal malformation was reported concerning a woman exposed to oseltamivir during the fourth month of pregnancy.

To our knowledge, only a few published studies have examined the effects of oseltamivir exposure during pregnancy. In 2011, a Swedish study compared 86 infants exposed to neuraminidase inhibitors during pregnancy (including only 24 women exposed during the first trimester), between 2005 and 2007, with 860 unexposed infants.[10] They did not identify any increased risk for adverse birth outcomes (SGA, low birthweight, preterm delivery, congenital malformation, and stillbirth) among the exposed group in comparison with the unexposed group. In 2010, an American study identified 137 women who received oseltamivir during pregnancy from the hospital pharmacy between October 2003 and March 2008, but only 18 women received the drug during the first trimester of pregnancy.[11] The study found no increase in preterm birth, stillbirth, and major or minor malformation rate in comparison with more than 80,000 controls. More recently, in 2013, a Canadian study compared 1237 women exposed to oseltamivir during the H1N1 epidemic with 54 118 unexposed women.[12] No evidence of an association between exposure to oseltamivir and adverse birth outcomes (SGA, preterm delivery, low Apgar score at birth) has been demonstrated. Our findings are in line with these reports.

Conclusion

The present study did not find any significant association between adverse fetal outcomes, including pregnancy loss for any cause, preterm delivery, low birthweight, neonatal pathology, congenital malformation, and exposure to oseltamivir during pregnancy. Our findings add reassurance about the use of oseltamivir during pregnancy. Nevertheless, further research is needed to increase the power and robustness of the analysis by including a larger number of exposed pregnant women.

Disclosure of interests

None of the authors have any potential conflicts of interest to declare.

Contribution to authorship

All authors contributed to writing the article. The study was planned by IL, CDM, and TV. Both IL and CDM supervised the study. CHD contributed to the data collection and statistical analysis. AB analysed the data and wrote the first draft of the article. IL and CDM helped to interpret the data and assisted in writing the article. TV and JLM critically reviewed the article for intellectual content. All authors read, commented, and approved the final version.

Details of ethics approval

The EFEMERIS cohort was approved by the French Data protection Agency on 7 April 2005 (authorisation number 05-1140).

Funding

The EFEMERIS database has been funded by the Agence Nationale de Sécurité du Médicament et des produits de santé (ANSM), the Caisse Nationale d'Assurance Maladie des travailleurs salariés (CNAMTS), the Mutuelle Générale de l'Education Nationale (MGEN), the Clinical Research Hospital Programme (PHRC), and the Unions régionales des Caisses d'Assurance Maladie (URCAM). Additional funding has been received from the ANSM for the study of influenza medication.

Acknowledgements

We are grateful to S Vidal, L Finotto, C Guitard, C Vayssière, and D Petiot for data collection.

Ancillary