Please cite this paper as: Källén B, Olausson P. Vaccination against H1N1 influenza with Pandemrix® during pregnancy and delivery outcome: a Swedish register study. BJOG 2012;119:1583–1590.
Objective To describe a large study on pregnancy outcome after vaccination against H1N1 during the 2009/10 pandemic.
Design A cohort study of women vaccinated with Pandemrix® during pregnancy.
Setting The Swedish Medical Birth Register was used for the analysis. Information on vaccination and pregnancy week when vaccination was made was obtained from antenatal care documents.
Population All women who gave birth during 2009 and 2010 in Sweden.
Methods Characteristics of the vaccinated women and their delivery outcome were compared with two groups of women: women without a known vaccination who gave birth in 2009/10 after 1 October 2009, and women who gave birth during 2009 before 1 October. Adjustment was made for year of delivery, maternal age, parity, smoking habits and body mass index.
Outcome measures Stillbirth, congenital malformations, preterm birth, low birthweight, small for gestational age.
Results A total of 18 612 vaccinated women having 18 844 infants were studied. The risk for stillbirth, preterm birth and low birthweight was lower than in the comparison groups whereas the risk for small for gestational age and a congenital malformation (after vaccination during the first trimester) did not differ from the comparison groups. No clear-cut explanation to the ‘protective’ effect of vaccination was found.
Conclusions Vaccination during pregnancy with Pandemrix® appeared to have no ill effects on the pregnancy. On the contrary, the rate of preterm birth and low birthweight was lower than expected, which agrees with some previous results.
The pandemic of H1N1 influenza, sometimes called swine influenza, began during April 2009 and reached Sweden the same year in early May. In October 2009, a mass vaccination against H1N1 influenza was started in Sweden. A vaccine produced by Glaxo Smith Kline (Brentford, Middlesex, UK) was used, Pandemrix®. This vaccine contained inactivated split influenza virus (A/California/07/2009), squalene adjuvant and thiomersale preservative.
The official recommendations regarding vaccination during pregnancy for H1N1 influenza varied between 20 different European countries and also the type of vaccine used.1 In 12 countries, the recommendation was to only vaccinate during the second or third trimester. In Sweden, pregnant women were included in the vaccination programme.
Until recently, studies on H1N1 influenza vaccination were restricted to small investigations based on only a few hundred women.2–5 An article in press6 followed 332 women who were vaccinated for H1N1 influenza and compared them with 1329 nonvaccinated women and found no significant difference in spontaneous abortion rate or the rate of congenital malformations. This conclusion was based on five spontaneous abortions and two infants with malformations born after vaccination before or during the first trimester.
A Danish study7 found that 12.9% (n = 7062) of pregnant women who were vaccinated against H1N1 influenza had no increase in spontaneous abortion rate and a lower-than-expected stillbirth rate.
A study in Ontario, Canada,8 identified 23 340 women who were vaccinated during pregnancy against H1N1 influenza and found that vaccinated women were less likely to have an small-for-gestational-age (SGA) infant (relative risk [RR] 0.90, 95% CI 0.85–0.96), preterm birth (RR 0.73, 95% CI 0.58–0.91) or fetal death (RR 0.66, 95% CI 0.47–0.91).
Many authors have stressed the need for clear-cut safety data concerning influenza vaccination during pregnancy.9–11
To study rare outcomes, a large number of exposed women are needed. We present a study based on national health registers from Sweden on delivery outcome after H1N1 vaccination and describe neonatal outcomes: the presence of congenital malformations, preterm birth, low birthweight, and intrauterine growth restriction.
Complete vaccination registers were only kept in some counties in Sweden. During antenatal care, however, information on vaccination was often obtained and recorded in standardised antenatal medical documents. In these documents, space is present for the recording of drugs and vaccinations taken since the woman became pregnant and before the first antenatal visit (usually during pregnancy weeks 10–12) and also during the subsequent antenatal care. The recording is probably incomplete but when a note exists that a vaccination has been made, it is likely to be correct. The week of pregnancy when the vaccination was made was generally recorded but this information was missing in 18% of the women.
These medical records were entered into the Medical Birth Register database.6 This register contains information on antenatal care, delivery and the neonatal condition of the newborn infant. Some information on the mother’s social situation is also included in the register. From this register, data on all women with a recorded H1N1 vaccination during pregnancy were extracted, a total of 18 623 women (vaccinated group). The pregnancy and delivery outcomes were compared with two groups of women, women delivered after 1 October 2009 up to the end of 2010 without information on vaccination (nonvaccinated group), and women delivered during 2009 before 1 October, that is, a period when no mass vaccination of the population occurred (prevaccination group). It should be stressed that within the nonvaccinated group an unknown number of women had been vaccinated but the vaccination had not been recorded. Some women in the prevaccination group may have had vaccinations against seasonal influenza. Such vaccinations have been recommended to pregnant women but not until after week 15.
From the Medical Birth Register,12 information on maternal age, parity, smoking habits in early pregnancy, prepregnancy body mass index (BMI) and woman’s country of birth were obtained. Also some maternal pregnancy diagnoses were included in the analysis and also gestational duration, usually determined by second-trimester sonography. Postdelivery information on stillbirth, birthweight and SGA infants, determined after normal graphs constructed from the register,13 were extracted.
Information on congenital malformations among the infants was present in the Medical Birth Register. More extensive information on congenital malformations were obtained by linkage with the Birth Defect Register (previously called the Congenital Malformation Register) and the Patient Register for 2009 and 2010, containing data on hospital discharge diagnoses and also information from specialist outpatient diagnoses. The use of different data sources will increase the malformation ascertainment but this will depend on the length of follow-up why comparisons were only made with children born in 2010. In this analysis, some specific groups of malformations were analysed: congenital cardiovascular defects (except single umbilical artery and patent ductus at preterm birth), ventricular and atrial septum defects, eye malformations, orofacial clefts and hypospadias. An analysis was also made after exclusion of some common and clinically less significant conditions, which also show a marked variability in registration: preauricular tags and sinuses and other branchial malformations, tongue tie, single umbilical artery, patent ductus at preterm birth, undescended testicle, congenital deformity of the sternocleidomastoid muscle, (sub)luxation of hip and nevus. The remaining malformations were categorised as ‘relatively severe’.
Characteristics of women who were vaccinated were compared with characteristics of nonvaccinated women and with women who belonged to the prevaccintion group (and the vast majority of which could therefore not have been vaccinated). The following maternal characteristics were studied: age (5-year classes, <20, 20–24 etc.,), parity (1, 2, 3, ≥4, where 1 means first child born), smoking in early pregnancy (unknown, none, <10 cigarettes/day, ≥10 cigarettes/day), BMI (unknown, <19.8, 19.8–24.9, 25–29.9, 30–39.9, ≥40), and if the woman was born in Sweden or outside Sweden. The presence of a delivery diagnosis of pre-existing diabetes, gestational diabetes or pre-eclampsia was also investigated.
Infant outcome referred to stillbirths, presence and types of congenital malformations, occurrence of preterm births (<37 weeks), low birthweight (<2500 g) and SGA in singletons when vaccination was performed before week 37. In the study of malformations, only vaccinations made before the end of the first trimester were included.
In further analyses, comparisons were made with nonvaccinated women who had reached the gestational week when vaccination was made in the vaccinated women, which of course could only be made when the vaccination week was known. Comparisons of the occurrence of preterm births were made between vaccinated women that week and nonvaccinated women who had not yet given birth.
Adjustments as shown in the tables were made using the Mantel–Haenszel method and approximate 95% confidence intervals were estimated with Miettinen’s method.
The total number of vaccinated women was 18 612 having 18 844 infants (vaccination group). These women were compared with 136 914 women having 138 931 infants who gave birth after September 2009 and before the end of 2010 (nonvaccinated group) and with 83 298 women having 84 484 infants who gave birth in the year 2009 before October (prevaccination group).
Time of vaccination for H1N1 (first immunisation)
About 18% of the vaccinations recorded in the Medical Birth Register had no information on the pregnancy week when the vaccination was made. The distribution of the remaining women is shown in Figure 1. There is a peak around the end of the first trimester (when the first antenatal care visit is usually made) and a second peak around 24 weeks of gestation.
Variation in the registration of vaccination between counties
Sweden is divided into 21 counties or regions. The percentage of women who had a recorded vaccination during pregnancy varied, being <1% (three counties), 1–14% (ten counties) and 15–22% (eight counties). Among women who were vaccinated and had a stated week of vaccination in the register, the percentage of first-trimester vaccinations varied from <10% (two counties), throught 10–19% (ten counties) to 20–29% (nine counties).
Table 1 shows some maternal characteristics, comparing vaccinated women and nonvaccinated women, and also vaccinated women with women who belonged to the prevaccination group. There were only slight differences between the two sets of odds ratios (OR) calculated for maternal characteristics.
Table 1. Numbers and characteristics of vaccinated women, nonvaccinated women and women belonging to the prevaccination group
No. in prevaccination group
Vaccinated versus nonvaccinated
Vaccinated versus prevaccination group
Adjusted odds ratios (OR) with 95% confidence intervals (95% CI) compare the three groups. The odds ratio for each variable is adjusted for all other variables in the table.
Maternal age (years)
The highest recorded vaccination rate was seen among women aged 25–29 years and among women of parity 2–3. Maternal smoking in early pregnancy did not significantly affect vaccination rate. There was an increased vaccination rate with increased BMI.
Women who were born outside Sweden were vaccinated much less often than Swedish-born women. The OR for being vaccinated when born outside Sweden was 0.63 (95% CI 0.60–0.65) when compared with nonvaccinated women and 0.74 (95% CI 0.71–0.77) when compared with women belonging to the prevaccination group.
Table 2 shows the effect of maternal morbidity on vaccination rate. Women with pre-existing diabetes showed a lower vaccination rate than other women. This effect was slightly stronger when the analysis was restricted to Swedish-born women. No effect was seen for gestational diabetes or pre-eclampsia. Both estimates for pre-existing hypertension and chronic lower respiratory disease were increased but the only estimate reaching statistical significance was for chronic lower respiratory disease when compared with the prevaccination deliveries.
Table 2. Number of specified delivery diagnoses among vaccinated women, nonvaccinated women, and women belonging to the prevaccination group
No. in pre-vaccination group
Vaccinated versus nonvaccinated
Vaccinated versus prevaccination group
Odds ratios (OR) with 95% confidence intervals (95% CI) for having each diagnosis among the three groups, adjusted for year of birth, maternal age, parity, smoking and BMI.
*Nearly identical ORs were obtained if women with pre-existing diabetes were excluded.
**Chronic lower respiratory disease, including asthma.
Five neonatal outcomes are summarised in Table 3. The stillbirth rate was lower after maternal vaccination than among the two control groups. This was also seen for the rate of preterm birth and low birthweight but not for SGA or for congenital malformations registered in the Medical Birth Register. When the analysis of congenital malformations was repeated restricted to vaccinations between weeks 1 and 9, still no increased risk was seen (OR 0.93, 95% CI 0.81–1.22, based on 54 malformed infants among 1729 exposed infants).
Table 3. Outcome among newborn infants
No. in pre-vaccination group
Vaccinated versus nonvaccinated
Vaccinated versus prevaccination group
Odds ratio (OR) with 95% confidence interval (95% CI) for outcome after maternal vaccination for H1N1 during pregnancy, adjusted for year of birth, maternal age, parity, smoking and BMI. Preterm birth, <37 weeks; low birthweight, <2500 g; SGA, <2 SD from expected weight at the relevant gestational week.
*Refers to singleton infants and vaccinations before 37 weeks of gestation.
**Any congenital malformation registered in the Medical Birth Register. Refers to vaccinations made during the first trimester (n = 3165). See also Table 5.
Stillbirth, only Swedish-born
Preterm birth, only Swedish-born*
Low birthweight, only Swedish-born*
SGA, only Swedish-born*
Congenital malformations, only Swedish-born**
When the analysis was repeated comparing vaccinated women with women who had given birth during 2009 but before 1 May (when influenza reached Sweden), a similar OR was obtained.
When an analysis was made restricted to women with known vaccination week and with comparisons with nonvaccinated women who were still pregnant in that week, all ORs increased and were no longer significantly <1.0. For stillbirths the OR = 0.95 (95% CI 0.69–1.22). for preterm birth among singletons OR = 0.96 (95% CI 0.88–1.05), for low birthweight in singletons OR = 0.96 (95% CI 0.87–1.07), and for SGA OR = 1.05 (95% CI 0.94–1.18). When the data were divided into three groups according to the timing of vaccination, an effect on the risk of preterm birth was seen only when vaccination was made after week 26 (Table 4). The preterm rates in the three groups differed significantly (chi-square for 2 df 13.8, P = 0.001).
Table 4. Effect of vaccination, when comparing vaccinated women with nonvaccinated women who had an ongoing pregnancy at the week of vaccination
Vaccination week of gestation
The effect on preterm births is analysed for three groups according to week of vaccination. Odds ratio (OR) with 95% confidence interval (95% CI) for preterm birth after maternal vaccination for H1N1 during pregnancy. Adjusted for year of birth, maternal age, parity, smoking and BMI.
In counties with a high registered vaccination rate (>10%), the OR for preterm birth was 0.84 (95% CI 0.76–0.94) and in counties with a low registered vaccination rate (<10%) the OR was 0.93 (95% CI 0.79–1.12).
Among all women who were vaccinated during pregnancy between weeks 27 and 36, 89% gave birth during the period November 2009 to February 2010. These 4 months show the highest rates of preterm birth in the population (around 5.4%, yearly average 4.9%) and when the rate of preterm birth among vaccinated women was studied with a restriction to these 4 months, the OR was even lower: 0.54 (95% CI 0.45–0.65) when compared with nonvaccinated women.
The diagnoses among the 201 malformed infants identified from the data ascertained from multiple health registers (see Methods) are shown in Supplementary material, Table S1 The distribution of malformations shows no unusual feature but there are five infants with orofacial clefts, the mothers of three of these children were vaccinated during weeks 7–9, which is when the face is formed, the other two during week 11. The mothers of the infants with oesophageal or ilial atresia were vaccinated during weeks 8–9, slightly late for a causal association.
Risk analyses for subgroups of malformations identified no statistically significant excess risk (Table 5). The highest estimate concerned ventricular and atrial septum defects but this estimate was also far from statistically significant.
Table 5. Risk for congenital malformations among infants whose mothers were vaccinated during the first trimester (n = 3197)
No. after vaccination
Malformations were identified from multiple sources. Comparisons with infants born of nonvaccinated women (n = 111 297). Odds ratios (OR) with 95% confidence intervals (95% CI) adjusted for maternal age, parity, smoking and BMI.
*As the expected number was low, the risk estimate is made as an observed over expected ratio with exact Poisson confidence interval.
Relatively severe malformations
Relatively severe malformations, vaccination before week 10
Any cardiovascular defect
Ventricular/atrial septal defect, no other cardiac defect
Most infants (84%, n = 2643) exposed in the first trimester to maternal Pandemrix vaccination were born during May to July 2010. The risk for a ‘relatively severe malformation’ among these infants was 0.88 (95% CI 0.72–1.09).
After exclusion from the analysis of women who were born outside Sweden, only small changes in OR estimates occurred but the low OR for stillbirths when compared with women belonging to the prevaccination group lost statistical significance.
This is one of the largest published studies of delivery outcome after vaccination against H1N1 during pregnancy. Most previous publications are based on much smaller samples. Among 130 pregnancies in a Scottish study, there were four miscarriages and six possible congenital malformations.2 A French study identified 569 pregnant women who were vaccinated with a nonadjuvant H1N1 vaccine (Panenza®, Sanofi Pasteur, Lyon, France) and found no adverse effects.3 A US prospective study of 261 pregnancies where vaccination for an AS03-adjuvant split virion H1N1 had been performed found a normal delivery outcome.4 A Taiwan study analysed 198 pregnancies (202 infants) where the women had been vaccinated with an adjuvant-free vaccine (AdimFlu-S®, Adimmune Corp., Tanzih Township, Taiwan) without finding any signs of adverse pregnancy effects.5
Only one vaccine had been used in our study, Pandemrix, and the results may not be relevant for other vaccines. Pandemrix contains both an adjuvant (squalene) and a preservative (thiomersale).
The advantage with this study is, in addition to the large number of women, the use of health registers which makes it possible to study characteristics of the vaccinated women and adjust for putative confounders, and to ascertain outcome in an unbiased way.
The study has not identified all vaccinated women. When comparisons are made with deliveries that occurred after the start of the mass vaccination (October 2009), some women who are regarded as nonvaccinated will in fact have been vaccinated. This will bias the estimated ORs towards unity, which was also seen when the OR for preterm birth was compared between counties with a high (OR = 0.84) and a low (OR = 0.93) registered vaccination rate. To avoid this fallacy, we also made comparisons with women who had given birth during 2009 before October and therefore were most certainly not vaccinated for H1N1 influenza during pregnancy. Some may have had vaccination for seasonal influenza, however. The estimated two ORs were in most instances similar even though a tendency can be seen that the ORs based on comparisons with ‘nonvaccinated’ women deviate slightly more from 1.0 than the ORs based on comparisons with women delivered before October 2009.
We found no effect on congenital malformation rate after vaccination and the low stillbirth rate after vaccination could be random but agrees with the results in a recent Danish study.7 The risk of preterm birth (and therefore also of low birthweight) is significantly lower than 1.0 after vaccination during pregnancy. This analysis was restricted to deliveries where the vaccination had been made before 37 weeks of gestation because if made later, it could not cause a preterm delivery. No effect was seen on intrauterine growth restriction or SGA. Further analyses showed that the effect of the vaccination on preterm birth was only seen when it was given after 26 weeks of gestation and that it was not the result of the temporal pattern of vaccinations, which could have confounded the analysis. A similar effect was seen in a study of maternal influenza immunisation (not against H1N1) with an OR of the same order of magnitude as that found by us.14
Different alternative explanations can be discussed for the finding of a better than expected delivery outcome after vaccination with regard to preterm birth and low birthweight. It could be an effect of uncontrolled confounding, for instance, from socio-economic conditions. It could be that a high socio-economic standard increased the probability of vaccination and decreased the probability of preterm birth. Two facts speak against this explanation. One is that practically no association was seen between maternal smoking and vaccination rate, and maternal smoking is in Sweden today strongly associated with a less advantageous socio-economic level. The second fact is that the effect on preterm birth was only seen when vaccination was performed after 26 weeks of gestation. Had the effect been a result of undetected confounding by socio-economic factors, one would expect similar effects of vaccination irrespective of the timing during pregnancy.
Another explanation of the ‘protective’ effect of vaccination could be that vaccination had been avoided in pregnancies showing pathological features, which could favour preterm birth. One such mechanism could be pre-eclampsia, but pre-eclampsia was not associated with a low vaccination rate. The only maternal morbidity that was associated with a reduced vaccination rate was pre-existing (but not gestational) diabetes. On the other hand, chronic lung disease was (as expected) associated with a somewhat increased vaccination rate but maternal asthma, if anything, increases the risk for preterm birth.15
Some women will have had influenza during pregnancy, mainly those who had not been vaccinated. If maternal influenza increased the risk for preterm birth, it could explain the significantly low risk after vaccination when comparison was made with nonvaccinated women who gave birth after September 2009. Also, among women who gave birth during 2009 before October 2009 influenza may have occurred. When the analysis was repeated comparing vaccinated women with women who had given birth before May 2009 (when the H1N1 influenza reached Sweden), the same low OR was found, however.
A direct effect of the vaccination must be considered. One possibility is that the immune reaction induced by the vaccination somehow reduced the risk for rejection of the fetus. A substantial part of noniatrogenic preterm deliveries are associated with signs of anti-fetal rejection.16 This would agree with the fact that a protective effect is only seen towards the end of the pregnancy (after 26 weeks of gestation) but this mechanism of action is speculative. Inflammatory responses to trivalent influenza vaccine among pregnant women have been described that were suggested to be beneficial for pregnancy outcome.17
The observed effect could, however, also be obtained if vaccination before 22 weeks of gestation increased the risk for spontaneous abortion in pregnancies that were at risk for preterm birth (and stillbirth), e.g. as the result of placental insufficiency. This is an unlikely explanation as vaccination before week 27 did not seem to affect preterm rates.
The number of women who were vaccinated before the end of the first trimester was only somewhat more than 3000 but among their infants no indication was noted of an increased risk for a congenital malformation. This observation is limited by the absence of data on prenatally detected and aborted fetuses. Such cases are registered but without identification number for legal reasons, and can therefore not be studied in relation to vaccinations. If vaccination caused a serious and nearly always detectable malformation (like anencephaly) this would not be found in this study, which is restricted to deliveries.
A possible teratogenic effect of vaccination may not act immediately at vaccination but could be delayed for some time if it was the result of immunological reactions. Vaccinations occurring during the latter part of the first trimester could then not cause malformations, because the embryo may then have left the window of organogenesis. An analysis of vaccinations made between 1 and 9 weeks of gestation did not change the risk estimate, however.
In conclusion, we found no ill effects of vaccination with Pandemrix during pregnancy but instead an apparent beneficial effect in a marked reduction of the rate of stillbirth, preterm birth and low birthweight, but no effect on SGA or malformation rate. Possible long-term effects were not studied because data are not yet available.
Disclosure of interest
None of the authors declare any competing interests.
Contribution to authorship
The study was planned jointly between the two authors. BK analysed the data and wrote the first manuscript draft. POO supervised the collection of data, and read and commented on the manuscript. Both authors approved the final version.
Details of ethics approval
The study was performed within the responsibilities of the National Board of Health and Welfare and therefore no ethical approval from outside ethical committees was needed.