To conduct enhanced surveillance for signals of teratogenesis following use of the neuraminidase inhibitors zanamivir and oseltamivir in the treatment or post-exposure prophylaxis of 2009 A/H1N1 influenza during pregnancy.
To conduct enhanced surveillance for signals of teratogenesis following use of the neuraminidase inhibitors zanamivir and oseltamivir in the treatment or post-exposure prophylaxis of 2009 A/H1N1 influenza during pregnancy.
Prospective cohort study, using national surveillance data collected by the UK Teratology Information Service (UKTIS) during the 2009 A/H1N1 pandemic.
Pregnant women who were reported to UKTIS by healthcare professionals seeking advice about exposure to zanamivir and oseltamivir or to other non-teratogenic drugs.
Pregnancy outcomes were collected for prospectively reported pregnancies exposed to zanamivir (n = 180) or oseltamivir (n = 27), and compared with a reference group of 575 prospectively reported pregnancies exposed to non-teratogenic drugs over the same period.
Rates of major congenital malformation, preterm delivery and low birth weight.
No significant differences in overall rates of major malformation in live-born infants [adjusted odds ratios (aOR): zanamivir 0.37 (95% confidence interval 0.02–2.70); oseltamivir aOR 0.81 (0.05, 14.15)], preterm delivery [aOR: zanamivir 0.95 (0.45, 1.89); oseltamivir aOR 1.68 (0.38, 5.38)] or low birth weight [aOR: zanamivir 0.94 (0.25, 2.90); oseltamivir aOR 4.12 (0.59, 17.99)] were observed following exposure at any gestation. No major malformations were reported in 37 zanamivir or eight oseltamivir first trimester exposures.
These surveillance data do not provide a signal that use of zanamivir or oseltamivir in pregnancy is associated with an increased risk of the adverse pregnancy outcomes studied but the data are too limited to state conclusively that there is no increase in risk.
Influenza infection in pregnancy has been associated with an increased risk of adverse maternal and fetal outcomes, including maternal death, miscarriage and stillbirth,[1, 2] and less conclusively with congenital malformation in the offspring. Pregnant women were therefore identified as a high risk group requiring early treatment or post-exposure antiviral prophylaxis with the neuraminidase inhibitors zanamivir or oseltamivir during the 2009 Influenza A/H1N1 pandemic.[1, 4]
Oseltamivir is administered orally and is readily absorbed from the gastrointestinal tract, whereas zanamivir is an inhaled preparation and as such has lower systemic and subsequent transplacental bioavailability. For this reason, zanamivir was recommended in the UK as the preferred antiviral during the 2009 pandemic for prophylaxis in the pregnant patient where active illness had not been confirmed, especially during the first trimester, when organogenesis occurs. Oseltamivir was advised where A/H1N1 infection was suspected or confirmed in pregnancy on the basis that zanamivir may be less effective in the treatment of systemic illness. Other countries opted for the use of oseltamivir for first line use for in both scenarios.
Data on the fetal safety of neuraminidase inhibitor use during pregnancy was very limited at the onset of the 2009 pandemic, particularly for zanamivir, and consisted mainly of incomplete information from unpublished sources. To address the need to collect information on maternal and fetal outcome during the 2009 pandemic, the UK Teratology Information Service (UKTIS) collected fetal outcome details from health care professionals, the vast majority of whom were general practitioners, who contacted the service for case-specific advice and information about the safety of oseltamivir or zanamivir use in a particular patient who was pregnant. In this report, these pregnancy outcome data are compared with information collected over the same period from enquiries relating to other drug treatments considered non-teratogenic.
Data collection was performed as surveillance on behalf of the Health Protection Agency and was covered by Section 251 of the NHS Act 2006. Because this data collection was conducted as enhanced surveillance activity by the UK Teratology Information Service during the pandemic, separate ethical approval was not required.
For part of the study period, UKTIS was also involved in a study funded by the National Institute of Health Research investigating the characteristics and management of women with or at risk of A/H1N1 virus infection and the relationship to pregnancy outcomes. The study was approved by the County Durham and Tees Valley Research Ethics Committee (REC reference number: 09/H0905/66). All participants provided informed consent. The study sponsor was the Newcastle upon-Tyne Hospitals NHS Foundation Trust (NUTH). Data collected during this study for women exposed to neuraminidase inhibitor antiviral agents were also included in the current analysis.
Information was collected for singleton pregnancies reported between January 2009 and December 2010 with outcome data collected until October 2012. Exposed pregnancies were defined as women who received the neuraminidase inhibitors zanamivir or oseltamivir in pregnancy. The reference group comprised women about whom UKTIS received enquiries during the same time period regarding exposure to other medications that are considered not to be teratogenic. These include topical preparations, cyclizine, amoxicillin and paracetamol. Women also exposed to confirmed teratogens (e.g. oral retinoids, sodium valproate, methotrexate and mycophenolate mofetil) were excluded from the comparison group. This method of comparison is internationally accepted among Teratogen Information Services.
Women in the neuraminidase inhibitor or reference groups were identified to UKTIS by healthcare professionals when clinical advice was sought from the service. Patient data was ascertained prospectively where the outcome of the pregnancy and the results of prenatal diagnosis were not already known at the time of the enquiry.
Details of exposures were collected at the initial contact with the service. Immediately following this, a questionnaire was sent to the enquirer requesting confirmation of information including maternal demographics, obstetric and medical history, social or illicit drug exposure and smoking status.
Details of pregnancy outcome were requested by a further questionnaire administered approximately 1 month after the expected date of delivery. These included details of any additional exposures, pregnancy complications and outcome, gestational age, birth weight, gender, APGAR scores, congenital malformations and neonatal problems. Up to two reminder letters and questionnaires were sent out to health professionals who did not respond.
The primary outcome analysed was the rate of major congenital malformation in live born singleton infants following first trimester exposure to a neuraminidase inhibitor. Rates of preterm delivery and low birth weight in term infants were also investigated for exposure at any stage of pregnancy. Rate of live birth, miscarriage, elective termination, late fetal death, and minor congenital malformation are also reported to contextualise the primary outcomes for these cohorts.
To compare apparent risk of specific pregnancy outcomes between the neuraminidase inhibitor and reference group, odds ratios (ORs) were calculated using logistic regression. Where a zero cell was present in a table, ORs were estimated using a zero-cell correction of 0.5 in the epiR package in r. Confidence intervals were calculated using exact methods and P < 0.05 was considered evidence for significant effect. Adjusted ORs were calculated for a reduced data set, adjusting for maternal age and week of exposure, where data were available. Data available were insufficient to allow for adjustment of other maternal factors that could potentially impact on the pregnancy outcomes studied, such as smoking, alcohol use, body mass index (BMI) and socioeconomic status.
Prematurity was defined as gestational age less than 37 weeks. Miscarriage was defined as the spontaneous loss of a pregnancy before 24 weeks' gestation. Late fetal death was defined as an intrauterine death or stillbirth at or after 24 weeks' gestation. Malformations were classified according to the EUROCAT classification system (revised July 2011) by an experienced observer who was blinded to maternal exposure status. The congenital anomaly rate includes live births only.
The outcomes of 782 pregnancies were available for analysis, including 180 exposed to zanamivir, 27 to oseltamivir and 575 exposed to other drugs. Of the 207 women exposed to neuraminidase inhibitors, 17 were recruited as part of the NIHR-funded study and the remainder via routine surveillance. Nine women included in the study also underwent vaccination for A/H1N1 influenza, six receiving zanamivir and three receiving oseltamivir. Details of maternal age, trimester of exposure, and indication for antiviral therapy for the study populations, are shown in Table 1.
|Mean maternal age (range)||30.2 (16–46 years)||27.5 (17–42 years)||28.9 (17–44 years)|
|Trimester of exposure|
|1||371 (64.5%)||40 (22.2%)||12 (44.4%)|
|2||136 (23.7%)||66 (36.7%)||9 (33.3%)|
|3||58 (10.1%)||69 (38.3%)||5 (18.5%)|
|Unknown||10 (1.7%)||5 (2.8%)||1 (3.7%)|
|Antiviral treatment||N/A||118 (65.6%)||16 (59.3%)|
|Antiviral prophylaxis||N/A||5 (2.8%)||0|
|Indication unknown||N/A||57 (31.6%)||11 (40.7%)|
Pregnancy outcomes are provided as Supporting Information (Table S1). Because of missing data, the number of observations reported for each outcome differs between the adjusted and unadjusted analysis, therefore the number of observations for each factor has been reported separately for each analysis.
The incidence of major malformation in first trimester neuraminidase inhibitor-exposed live-born infants was not significantly higher than that in the reference group (all first trimester neuraminidase-exposed 0/45 [0.0%] versus reference 4/281 [1.4%], aOR 0.73, 95% confidence interval [CI] 0.04–13.30; zanamivir 0/34 [0.0%] versus reference 4/281 [1.4%], aOR 0.89, 95% CI 0.05–16.96; oseltamivir 0/8 [0.0%] versus reference 4/281 [1.4%], aOR 3.63, 95% CI 0.18–72.89). Overall major and minor malformation rates were also not increased in live-born infants exposed at any stage of pregnancy (Table S1). The single reported major malformation (absent right kidney) occurred in an infant exposed to zanamivir at 27 weeks' gestation. This timing of exposure makes causality implausible if this is a case of true renal agenesis, but does not exclude the possibility of renal regression as a consequence of the exposure. The four minor malformations reported were a small paraumbilical hernia, positional talipes, hydronephrosis and spina bifida occulta, all in infants exposed to zanamivir during the second or third trimesters, with no malformations reported in the 45 live-born infants exposed to zanamivir or oseltamivir during the first trimester. A fetal cloacal abnormality was detected on antenatal scan following maternal treatment of suspected A/H1N1 infection with oseltamivir from day 10 to day 15 of pregnancy (periconceptually). The pregnancy was terminated electively at 23 weeks. Postmortem examination was declined and no further details are available. Exposure to external agents in the weeks prior to implantation (often referred to as the ‘all-or-nothing’ period) is thought unlikely to result in major congenital malformation in the fetus.
There were three (3.3%) elective terminations of pregnancy in the zanamivir-exposed group, two (11.1%) in the oseltamivir-exposed group including the case described, and 30 (6.6%) in the reference group. Other than the cloacal abnormality described above, none was known to have a diagnosis of fetal malformation.
Three miscarriages were reported in the neuraminidase-exposed groups, all to women who had been treated for influenza. These occurred (i) at 7 weeks' gestation, to a 24-year-old woman treated with oseltamivir at 4 weeks' gestation; (ii) at 9 weeks gestation to a 44-year-old woman with depression and a BMI of 31 kg/m2, treated with oseltamivir for influenza symptoms at 5 weeks' gestation. No other medication use was reported; (iii) at 21 weeks' gestation, 6 weeks after zanamivir treatment in a 22-year-old woman with asthma who was a smoker. Statistical comparison of the miscarriage rates was not undertaken, as the data was underpowered to perform a left truncated Cox analysis.
Gestational age at delivery was reported for 168 zanamivir-exposed, 20 oseltamivir-exposed and 445 unexposed live-born infants. Preterm delivery (<37 gestational weeks) occurred in 14 (8.3%) zanamivir-exposed infants, three (15.0%) oseltamivir-exposed infants and 46 (10.3%) infants in the reference group, with no significant differences between the groups (zanamivir aOR 0.95, 95% CI 0.45–1.89, oseltamivir aOR 1.68, 95% CI 0.38–5.38). One infant was delivered by emergency caesarean at 30 weeks' gestation. Hydrops fetalis had been detected and investigated antenatally, with a final diagnosis of non-isoimmune haemolytic disease. Placental abnormalities suggestive of chorioangioma were also observed; however, autopsy was not performed. The 36-year-old mother had tested positive for A/H1N1 and had been treated with zanamivir at 25 weeks' gestation. The baby subsequently died in the neonatal period. Reported maternal co-morbidities were gestational diabetes mellitus, polycystic ovary syndrome, chronic pain, vitamin D deficiency, endometriosis and irritable bladder. It is noteworthy that her two previous children had required neonatal blood transfusion in the neonatal period.
Birth weight was reported for 150 live-born infants exposed to neuraminidase inhibitors and 368 live-born infants in the reference group. Low birth weight (<2500 g) at term affected four (2.9%) zanamivir-exposed, two (14.3%) oseltamivir-exposed and 16 (4.3%) reference group infants, with no significant differences in incidence between the groups (zanamivir aOR 0.94, 95% CI 0.25–2.90, oseltamivir aOR 4.12, 95% CI 0.59–17.99).
Influenza infection in pregnancy has been associated with an increased risk of maternal mortality and morbidity, as well as spontaneous abortion, stillbirth and, less conclusively, certain congenital malformations in the offspring.[1-3] Several studies have demonstrated that in women with influenza, prompt antiviral use reduces maternal mortality and adverse pregnancy outcomes,[8, 9] although a small study using health registry data demonstrated an increased risk of late transient hypoglycaemia in infants exposed in utero. Early antiviral treatment of pregnant women with suspected influenza is therefore likely to remain a priority in future pandemics. The efficacy of neuraminidase inhibitor use for postexposure prophylaxis, and specifically for pregnant women remains uncertain but is likely to be recommended in pandemic settings, particularly for non-vaccinated women, given the risks associated with influenza infection in pregnancy.
Information on efficacy and adverse fetal effects of antiviral drugs, especially when used in the first trimester, is essential to enable women to make informed decisions regarding their use. In contrast to the high number of publications reporting on fetal outcome following vaccination against Influenza A/H1N1 in pregnancy, less surveillance data has emerged from the 2009 pandemic regarding the potential adverse fetal effects of neuraminidase inhibitors, particularly for zanamivir, which was recommended in the UK as the preferred choice for postexposure prophylaxis in pregnant women.
The largest study to date of neuraminidase inhibitor use in pregnancy during the 2009 A/H1N1 pandemic included 1237 pregnancies exposed to oseltamivir. There were no congenital malformations reported and no increase in the risk of small-for-gestational age, preterm birth, very preterm birth, or low Apgar score was demonstrated in the data set.
A prospective cohort of neuraminidase inhibitor-exposed pregnancies included 50 pregnant women exposed to zanamivir (15 in the first trimester) and 619 pregnant women exposed to oseltamivir (159 during the first trimester). No malformations were observed in the zanamivir-exposed infants. Two malformations (one cardiac defect, one haemangioma) were observed in the infants exposed to oseltamivir during the first trimester. The authors reported that these rates, and those for other pregnancy outcomes, did not differ from the national averages.
Prior to this, much of the published literature relating to oseltamivir use in pregnancy during the 2009 pandemic focused on treatment of severe A/H1N1 infection in later pregnancy and subsequent maternal outcome, with no additional information provided about the offspring beyond neonatal survival.[1, 4, 8] Two small cohort studies analysed fetal outcomes of women exposed to a neuraminidase inhibitor at any stage of pregnancy.[10, 13] Most exposures involved oseltamivir, with only two women receiving zanamivir as monotherapy and three women who received both. For zanamivir there is a single report involving three pregnancies where fetal outcome was reported for each individual exposure. Data regarding exposure during the first trimester (the period of risk for structural congenital malformation) or of fetal outcome following use of a neuraminidase inhibitor at any stage of pregnancy in the absence of maternal influenza infection, has been extremely limited.
The evidence presented here, including prospectively collected outcome data from 180 zanamivir-exposed pregnancies, does not provide any signal that use of zanamivir in pregnancy is associated with an increased risk of adverse fetal outcomes when compared with other pregnancies reported to UKTIS during the same period. It should be noted, however, that due to the nature of teratology information services, there is a propensity for enquiries to be made about higher risk pregnancies. However, overall malformation rate did not differ between groups and was no greater than the expected background population rate of major (2–3%) or major and minor (3–5%) malformation. Furthermore, there were no malformations reported in the 45 live-born infants exposed to a neuraminidase inhibitor during the first trimester.
Rates of spontaneous abortion and elective termination of pregnancy were low. The discrepancy in the number of first trimester exposures between the groups relates to the nature of the telephone enquiries. During the 2009 influenza A/H1N1 pandemic, advice was most often sought regarding antiviral exposure in the second and third trimester, where the risks due to influenza are greatest. In the reference group, however, enquiries to UKTIS are common during the first trimester, when advice is sought on the teratogenic effects of medications during organogenesis.
Rates of preterm delivery did not differ across cohorts and were similar to the reported overall UK rate of 7.0%. Similarly, the incidence of low birth weight singleton infants in the cohort was not significantly higher than the expected rate in the UK of 5.7%. The lack of adverse outcomes in the small number of oseltamivir-exposed pregnancies (n = 27) in this cohort are consistent with previously published human data that have shown no increased risk of miscarriage, elective termination, preterm delivery or congenital malformation compared with expected background rates.[9, 10, 13, 14, 17]
The strengths and limitations of observational cohort data collected by teratology information services are well documented. Although such studies generally involve a small sample size, the data collected are detailed, exposure information is accurate and they are often the only source of pregnancy safety data. These therefore remain an important method of teratogenic signal detection. However, the sample size is smaller than that required to assess accurately the outcomes studied, as is often the case with pregnancy surveillance data, which rely on observation of an incidental cohort. Our study provides the first prospective human pregnancy data for zanamivir and is strengthened by an internal comparison group, with detailed follow-up requested from healthcare professionals using consistent methodology. The limitations of our study include a small sample size, particularly for first trimester exposures and for oseltamivir, which reduces statistical power to detect increased frequencies of adverse outcomes, especially those that are uncommon. Furthermore, fetal outcome was only followed up around birth and anomalies of internal organs may therefore not yet have been detected, which may have contributed to the relatively low rate of anomalies reported amongst these cohorts. The observational design is potentially subject to confounding and bias, and details of some potential confounding factors were poorly reported by health care providers, precluding statistical adjustment. These included maternal smoking, underlying illness and severity, and concomitant medication use. Zanamivir was preferred in the UK for prophylaxis; however, most women in our cohort received it for treatment of suspected influenza, with only 2.8% receiving it as prophylaxis, preventing separate analysis of outcomes for those women with influenza infection.
This study, together with the currently available literature, does not provide a signal of major teratogenic or other adverse fetal outcomes from in utero exposure to zanamivir or oseltamivir, although statistical power to exclude such effects is limited. Benefits of treatment are therefore likely to exceed risks in pregnant women who are hospitalised with suspected influenza, as available data suggest clear maternal and fetal benefit from early neuraminidase inhibitor treatment. Less evidence is available of benefit from neuraminidase inhibitors in women with mild influenza symptoms or those receiving postexposure prophylaxis, but no specific fetal risks have been identified in this and other studies.[9, 10, 13-15] If experience of pandemic situations is to be used to inform future emergency planning, more effective methods of real time surveillance are needed, particularly for high risk and unique groups, including women who are pregnant.
The authors report no conflict of interest.
HJD led the preparation of the article, including writing and editing. ACM provided the statistical analysis. SS and LMY conceived the study and contributed to the writing and editing of the article. SHLT provided clinical guidance and contributed to the writing and editing of the article.
Data collection was performed as surveillance on behalf of the Health Protection Agency (covered by Section 251 of the NHS Act 2006), therefore separate ethical approval was not required. Data from the National Institute of Health Research study was approved by the County Durham and Tees Valley Research Ethics Committee (7 September 2009, REC reference number: 09/H0905/66). All participants provided informed consent.
Data collected for a study funded by the National Institute of Health Research was also included in this analysis.
We would like to acknowledge the support of the National Institute of Health Research for funding collection of some of the data used in this study and the Northumberland Tyne and Wear Comprehensive Local Research Network for invaluable assistance in carrying out that component of the study. We would also like to thank Prof. Judith Rankin of the Northern Congenital Abnormality Survey for interpretation of congenital malformation information. Finally we would like to thank all women and healthcare professionals providing information for this study.