The 2009-2010 A/H1N1 pandemic provided a unique setting to study the safety of MF59-adjuvanted vaccination in pregnancy.
This was an observational cohort study of the safety of an MF59-adjuvanted A/H1N1 vaccine (Focetria) conducted among 4508 pregnant women (2295 vaccinated vs 2213 unvaccinated), with 3 month follow-up of neonates.
No maternal deaths or abortions occurred among the vaccinated women. No differences between the vaccinated and unvaccinated cohorts were observed for gestational diabetes, preeclampsia, stillbirth, low birthweight, neonatal deaths, or congenital malformations. The risk of premature birth was significantly decreased among the vaccinated women (adjusted proportional hazard, 0.69; 95% confidence interval, 0.51–0.92). No differences were observed in rates of congenital malformations after vaccination in the first (2.1%), second (2.7%), or third (2.1%) trimesters.
There was no evidence of a safety risk for MF59-adjuvanted A/H1N1 vaccination in pregnant women; protection was observed against premature birth.
Soon after the outbreak of the A/H1N1 influenza pandemic in 2009, pregnant women were recognized as a high-risk group. In the United States, pregnant women had a 4-fold greater risk of hospitalization because of A/H1N1 infection than the general population at the start of the pandemic, and in the first months of the pandemic, 13% of all A/H1N1-related deaths reported in the United States occurred among pregnant women. Anticipated effects of A/H1N1 infection on the health of newborns were confirmed in the United Kingdom, where the risks of perinatal mortality as well as premature birth were 4-fold higher in mothers infected with A/H1N1 during their pregnancy.
See related editorial, page 145
For Editors’ Commentary, see Contents
Pregnancy has previously been recognized as a risk factor for complications both during influenza pandemics and regular influenza seasons. Given the good immune responses elicited by influenza vaccination in pregnant women and their offspring, vaccination has been recommended since the late 1950s to reduce the burden of seasonal influenza in pregnant women and their offspring. A recent randomized controlled interventional trial in Bangladesh showed a significant reduction in the influenza burden in both vaccinated pregnant women and their offspring. The protective effect in newborns was also shown in several observational studies.
Focetria (Novartis Vaccines and Diagnostics, Cambridge, MA), an egg-derived A/H1N1 influenza vaccine adjuvanted with MF59, was one of the first A/H1N1 vaccines to be licensed in Europe. MF59 is a squalene-based oil-in-water emulsion used in Europe since 1997 in the seasonal influenza vaccine, Fluad (Novartis Vaccines and Diagnostics). More than 50 million doses have been distributed to date, with no safety signals other than some increased local injection site reactions and general reactions associated with vaccination (myalgia, headache, fatigue, and malaise) up to 7 days after the vaccination.
A comprehensive preclinical assessment (including embryofetal toxicity and teratogenicity studies) and clinical trials in more than 20,000 subjects did not show any safety signal with the MF59 adjuvant. An analysis of the outcomes among 103 inadvertently vaccinated pregnant women in these trials did not find any adverse effect on the pregnancy outcomes. To further assess the safety of the MF59 adjuvant in pregnancy, we conducted an observational study to evaluate outcomes in pregnant women who received the MF59-adjuvanted A/H1N1 influenza vaccine during the recent pandemic.
Materials and Methods
This was an observational comparative cohort study conducted in 3 countries in which the MF59-adjuvanted A/H1N1 vaccine was the only vaccine administered to pregnant women: The Netherlands, Italy, and Argentina.
Recruitment and follow-up
There was no distinction in the way vaccinated women and controls were recruited. In The Netherlands, women were recruited at 27 midwife practices (76.0%) and 7 hospitals (23.4%) across the country. A small group (0.6%) was recruited by general practitioners. In Italy and Argentina, women were recruited at 2 hospitals (Rome and Cordoba). Recruitment ran from January to August 2010 in The Netherlands, May to June 2010 in Italy, and July to August 2010 in Argentina. To maximize the number of subjects, enrolment included both currently pregnant women (enrolled during antenatal care visits) and women whose pregnancy had already ended but covered the timing of the pandemic vaccine’s availability. The latter group was identified from the delivery list in the midwife or hospital unit and then recruited sequentially. All women who provided informed consent were enrolled except those who had received a different pandemic influenza vaccine or if the investigator believed that data collection or follow-up would be difficult.
Investigators collected information on demographics (date of birth, ethnicity, education, occupation, and income), previous obstetric history and past or current risk factors (tobacco, alcohol and recreational drug use, concurrent medications), results of any prenatal testing and the pregnancy outcomes, including all birth characteristics, at the prenatal consultations or at birth for the women enrolled while still pregnant. For women enrolled after delivery, this information was extracted from the medical records. Information from the pediatric 3 month visit was requested to identify any congenital malformations not detected at birth.
All data were collected on site via electronic data capture. Serious adverse events (SAEs) were also collected separately throughout the study and reconciled with the study outcomes, thereby providing an additional level of data verification for most outcomes of interest. Source documentation was verified for all study outcomes.
A/H1N1 vaccination of pregnant women occurred in November and December 2009 in The Netherlands, between October and December 2009 in Italy, and between February and August 2010 in Argentina. Each woman was asked following enrollment whether and when she had received the MF59-adjuvanted A/H1N1 vaccine. This information was confirmed by either asking the vaccination centers to confirm the vaccination from their records or by examining the vaccination cards. If no such confirmation could be obtained, we followed the pregnant woman’s recollection.
The 3 groups of outcomes of interest were pregnancy-related diseases (preeclampsia, gestational diabetes, and maternal death), pregnancy outcomes (spontaneous or induced abortions, stillbirth, or live birth), and birth outcomes (birthweight, prematurity, congenital malformations, and neonatal death). Abortions were considered as losses before 22 weeks of gestation. Congenital malformations were considered only if confirmed following examination of an aborted pregnancy or stillborn child or in a live-born baby at birth or at the 3 month pediatric visit. Each congenital malformation was also reported as an SAE, which included a detailed narrative.
These reports were reviewed and adjudicated by an expert panel, blinded to the vaccination status. Malformations were retained if the diagnosis provided is listed in the European Surveillance of Congenital Anomalies (EUROCAT) guidelines as a congenital anomaly or if the information provided sufficed to so classify it as such. Cases with a malformation with insufficient information to determine whether it met EUROCAT criteria (eg, hemangioma with unknown location) were conservatively retained. Malformations diagnosed prenatally before vaccination and chromosomal malformations were excluded from the analyses. Outcomes were recorded from time of vaccination in the vaccinated cohort and from the time of enrollment in the unvaccinated cohort recruited during pregnancy and from the whole antenatal care record for the unvaccinated cohort recruited after delivery.
A 2% background rate of major birth defects, as observed in The Netherlands, was assumed for the sample size calculation. Based on a 2 group χ 2 test with normal approximation at the 5% significance level (2 sided), 2434 evaluable pregnancies would provide 80% power to detect a doubling of the background rate (ie, anticipated rate in unvaccinated cohort). To allow for approximately 15% of pregnancies being nonevaluable for the assessment of major birth defects and assuming a 25% loss to follow-up, 4056 subjects needed to be enrolled.
All analyses were performed using SAS version 9.1 (SAS Institute, Inc., Cary, NC). We used logistic regression models to estimate the odds ratios associated with vaccination for all outcomes. The differences in initiation points for the data collection combined with the timing of the vaccination in pregnancy resulted in a differential follow-up time between the vaccinated and unvaccinated cohort. To account for this differential follow-up time and for the fact that the occurrence of most outcomes depends on the gestational age, we also used proportional hazard (PH) models with gestational age as the time factor for all outcomes except congenital malformations. For these PH models, we imputed the date of the last menstrual period (LMP) from the date of delivery for 10% of women who did not report their LMP and for 7% for whom the reported LMP was considered erroneous as it suggested a pregnancy of more than 42 weeks.
We provide results of both the logistic regression models that use the most complete data and the proportional hazard models that adjust for the differential follow-up time. We also ran the PH models without the women with missing or erroneous LMPs in a sensitivity analysis to assess the potential impact of the LMP imputation. A small proportion of women dropped out from the PH models because of missing vaccination dates (6%) or missing onset dates of the outcome (1%). We adjusted the estimates for all outcomes on parity, smoking, and maternal age.
In addition, we let the automated SAS procedure Stepwise select, per individual outcome of interest, any additional significant variables including enrollment type (during or after pregnancy), type of health care practitioner enrolling, ethnicity, current alcohol use, and previous history of the specific outcome of interest. Other variables were not included because they had too many missing data (education, profession, and income) or because they had very few positive entries (recreational drug use and concomitant medication).
The study was approved by local ethical review committees in each individual country.
Enrollment and follow-up
Of the total 4529 enrolled women, most (4281, 94.5%) were recruited in The Netherlands, followed by Argentina (239, 5.3%) and Italy (9, 0.2%). Twenty-one were excluded from the analyses (7 pregnancies completed prior to the vaccination campaign, 5 received a vaccine other than the MF59-adjuvanted A/H1N1 vaccine, 7 with incomplete or withdrawn consent, and 2 with unknown vaccination status). Because 16 women dropped out before pregnancy completion and 48 had twins, we had a total of 4540 evaluable pregnancy outcomes, of which 4522 were live births. We obtained information on the 3 month pediatric follow-up on 4385 infants (97.0%) ( Figure ).
Characteristics of the cohorts
The mean age at enrollment was 32 years and most women (83.3%) were of white origin. The vaccinated cohort was comparable with the unvaccinated cohort for most characteristics ( Table 1 ), with the exception that more vaccinated women were enrolled after delivery, belong to the “other” ethnicity or were enrolled by the “other” professional category and less vaccinated women had a previous history of an elective abortion.
|Characteristics||Vaccinated (n = 2295)||Unvaccinated (n = 2213)||Total (n = 4508)|
|Maternal age, y|
|Ethnic origin, n (%)|
|Asian||32 (1.4)||31 (1.4)||63 (1.4)|
|Black||16 (0.7)||21 (0.9)||37 (0.8)|
|White||1866 (81.3)||1890 (85.4)||3756 (83.3)|
|Hispanic||9 (0.4)||10 (0.5)||19 (0.4)|
|Other||372 (16.2)||261 (11.8)||633 (14.0)|
|Parity, n (%)|
|0||818 (35.6)||858 (38.8)||1676 (37.2)|
|1||769 (33.5)||696 (31.5)||1465 (32.5)|
|≥2||708 (30.9)||659 (29.7)||1367 (30.3)|
|Prior pregnancy complication, n (%) a|
|Spontaneous abortion||545 (36.9)||497 (36.7)||1042 (36.8)|
|Elective abortion||101 (6.8)||127 (9.4)||228 (8.1)|
|Stillbirth||22 (1.5)||15 (1.1)||37 (1.3)|
|Gestational diabetes||12 (0.8)||8 (0.6)||20 (0.7)|
|Preeclampsia||37 (2.5)||46 (3.4)||83 (2.9)|
|Premature delivery||85 (5.8)||87 (6.4)||172 (6.1)|
|Low birthweight||53 (3.6)||58 (4.3)||111 (3.9)|
|Neonatal death||8 (0.5)||7 (0.5)||15 (0.5)|
|Congenital malformation||45 (3.0)||37 (2.7)||82 (2.9)|
|Smoking, n (%)|
|Never||1470 (64.1)||1371 (62.0)||2841 (63.0)|
|Current||166 (7.2)||221 (10.0)||387 (8.6)|
|Ex-smoker||308 (13.4)||339 (15.3)||647 (14.4)|
|Missing||351 (15.3)||282 (12.7)||633 (14.0)|
|Current alcohol consumption, n (%)|
|Yes||58 (2.5)||49 (2.2)||107 (2.4)|
|No||2056 (89.6)||2032 (91.8)||4088 (90.7)|
|Missing||181 (7.9)||132 (6.0)||313 (6.9)|
|Current recreational drug use, n (%)|
|Yes||3 (0.1)||6 (0.3)||9 (0.2)|
|No||2151 (93.7)||2077 (93.9)||4228 (93.8)|
|Missing||141 (6.1)||130 (5.9)||271 (6.0)|
|Timing of enrollment, n (%)|
|During pregnancy||793 (34.6)||1289 (58.2)||2082 (46.2)|
|Gestational age at enrollment, weeks|
|After delivery||1502 (65.4)||924 (41.8)||2426 (53.8)|
|Timing first vaccination, n (%)|
|First trimester||92 (4.0)||n/a||n/a|
|Second trimester||1307 (56.9)||n/a||n/a|
|Third trimester||888 (38.7)||n/a||n/a|
|Missing b||8 (0.3)||n/a||n/a|
|Doses of MF59-adjuvanted H1N1 vaccine received, n (%)|
|One dose||571 (24.9)||n/a||n/a|
|Two doses||1724 (75.1)||n/a||n/a|
|General practitioner||10 (0.4)||15 (0.7)||25 (0.6)|
|Obstetrician||483 (21.0)||524 (23.7)||1007 (22.3)|
|Midwife||1603 (69.8)||1634 (73.8)||3237 (71.8)|
|Other||199 (8.7)||40 (1.8)||239 (5.3)|
Pregnancy and neonatal outcomes
There were no maternal deaths. Nearly all pregnancies (99.6 %) resulted in a live birth, a rate mostly explained by the relatively high gestational age at enrollment in the cohort enrolled during pregnancy. For the same reason, there were few spontaneous abortions and elective terminations (0.2% and 0.1%, respectively). The rates of prematurity and low birthweight were relatively low compared with rates in The Netherlands where most women were recruited, suggesting a relatively healthy cohort. Potential congenital malformations were reported in a total of 365 babies: 188 (8.1%) among the vaccinated and 177 (7.9%) among the unvaccinated. Of these, the adjudication committee retained 106 as true congenital malformations, 9 of which were excluded from the analysis (2 were diagnosed prenatally before vaccination and 7 were chromosomal malformations). Thus, the total number of confirmed congenital malformations included in the analysis was 97.
The logistic regression and proportional hazard analyses did not indicate a significantly increased risk in the vaccinated compared with the unvaccinated cohort for any of the outcomes of interest. Vaccination was not associated with any significant increase in risk of gestational diabetes, preeclampsia, stillbirth, low birthweight, preterm birth, neonatal deaths, or congenital malformations ( Table 2 ). In fact, there was a significant reduction in the vaccinated cohort in premature births in the unadjusted logistic regression and both the adjusted and unadjusted PH models.