Our results show that antibodies against H1N1 2009 pandemic virus are present in cord-blood samples from substantially more vaccinated than unvaccinated women and at much higher titres. Our findings are particularly relevant given the substantial impact of H1N1 2009 virus on pregnant women 
and the high hospitalization rates and increased morbidity among young infants. 
However, these results cannot be applied to babies born before 34 weeks gestation as only babies born after this point were recruited. It is probable that the findings would differ for more premature babies as although active antibody (IgG) transfer begins from about 13–16 weeks gestation it does not reach a peak until the last 4 weeks of pregnancy. 
In our study over 90% of cord-bloods from vaccinated mothers had HI titres of ≥1
8 and MN titres of ≥1
10. These high sero-prevalence rates (>90%) contrast with levels of 37% for the presence of HI and MN antibodies in cord-bloods from non-vaccinated women. This ‘background rate’ of 37% for both HI and MN antibody in cord blood from non-vaccinated mothers is comparable to sero-incidence rates of 10–40% found in venous blood collected across England from people aged 25- to 44-years during October 2009–February 2010. 
Although the baseline characteristics of vaccinated and unvaccinated women were generally similar, cord-bloods from vaccinated women were collected a median of 9.5 days after those from unvaccinated women. Multivariable logistic regression modelling showed that duration of exposure had no significant effect on cord antibody findings. In the UK, the majority of women enter secondary care for parturition. The three hospitals where participants were recruited provide secondary and tertiary antenatal and intra-partum obstetric care to expectant women resident in the catchment area around them. It is therefore unlikely that any referral/enrolment bias has occurred. In all, we therefore conclude that the substantial difference in sero-prevalence rates (>50%) was the result of maternal vaccination rather than natural infection of vaccinees.
Although there are no established correlates of protection for HI or MN in children and infants, evidence is accumulating that antibodies against seasonal influenza viruses, at levels that are normally associated with protection, can be found in a high proportion of cord blood samples 
and neonates 
following immunization during pregnancy. Our study found that three-quarters of vaccinated women had ‘seroprotective’ cord blood HI antibody titres of ≥1
40 that are associated with at least 50% protection against symptomatic influenza in adults. 
The proportion with MN titres above 1
≥60 were higher than the proportion achieving protective HI titres (84.2% versus 75.3%). This pattern is consistent with the findings of Veguilla et al 
, who found that MN titres were more sensitive but HI titres more specific (for natural infection). Correlates of protection for MN titres are less clear than for HI titres and therefore the choice of cut-off for MN titres was arbitrarily chosen as ≥1
60. Other studies have used thresholds ranging from 1
40 to 1
80 and therefore the midpoint was used in this study 
. As the primary endpoint was HI titres, and MN titres ≥1
320 add little further information, endpoint titration for MN analysis was not performed. The MN titres in babies of vaccinated mothers were 12 fold higher than in those from unvaccinated mothers. This may however be an underestimate because endpoint titration MN titres were not available.
There is evidence that maternal immunization with seasonal influenza vaccine prevents laboratory-confirmed influenza in infants, 
and hospitalizations, 
so our finding of HI and MN antibodies in cord-bloods from vaccinated mothers, particularly ‘seroprotective’ HI titres of ≥1
40, is likely to be beneficial during the first few months of life when influenza attack rates and illness severity are high and preventative and therapeutic options for infants are limited.
A key finding in our study was the appearance of antibody in cord blood shortly after maternal immunization. By day 8, 97.1% and 95.5% of cord bloods respectively contained HI and MN antibody against pandemic H1N1 2009 virus, 77.9% of cord bloods had ‘seroprotective’ HI titres of ≥1
40 and 86.6% had MN titres of ≥1
60. The trend-line fitted to the scatter-chart of HI antibody titres in relation to time between vaccination and delivery () was consistent with passive transfer of protective levels of antibody by day 7 after maternal immunization. Our cord blood antibody results are consistent with those seen in venous blood one week after immunization of young adults and healthcare workers with AS03A
-adjuvanted H1N1 2009 vaccine. 
Our study suggests that vaccination of pregnant women with AS03A
-adjuvanted H1N1 2009 vaccine as late as one week before the expected date of delivery could benefit both mother and child, even if administered during a localized outbreak. We found no evidence of an age-related decline in either HI or MN antibody levels in pregnant women when assessed in age bands 16 to 29 years and 30 to 44 years. Thus children born to 30 to 44 year-old women seem just as likely to benefit from maternal immunization with AS03A
-adjuvanted H1N1 2009 vaccine as those born to younger women. It is also evident that vaccination stimulates a sustained maternal response for up to 14–16 weeks; this may have been influenced in part by any natural exposure to antigenically-related influenza virus. (). Median titres remain above protective levels to this point, however, it is important to note that the range is wide (possibly because few data points were available at this extreme duration) and that for the first quartile, titres dropped below the protective level by 12–14 weeks. Additional research is needed to elucidate for how long the duration of protection extends. However, the rate of decline of actively induced antibody in mothers may be different to the rate of decline of passively acquired antibody in their infants. Antibody levels in the infant will decline after birth; the precise rate of decline is unclear but may be influenced by organism type, starting levels and the mode of immunity (natural vs. vaccinated) 
. Ochola et al demonstrated the half-life of passively acquired maternal RSV antibodies in infants was 79 days. 
Other studies of the persistence of passive maternal antibodies to H.pylori and measles both suggest that immunity has waned by 6 months after birth 
; however other work suggests the half-life period for influenza antibodies may vary. 
We found no evidence for an effect of other potential confounders on the observed antibody responses. It should be noted that we only studied cord antibody levels in response to AS03A
-adjuvanted H1N1 2009 vaccine, so it is unclear whether conventional (i.e., unadjuvanted) split and subunit vaccines, whole virus vaccine, or vaccine with alternative adjuvants trigger H1N1 2009 HI and MN antibody as rapidly in cord blood as in our study, or to the same high titres. (Although two brands of vaccine were available for use in the UK during the pandemic, only the AS03A
-adjuvanted H1N1 2009 vaccine was recommended for use in pregnant women). 
Pandemic influenza vaccines containing the MF-59 (oil in water emulsion) adjuvant were not procured in the UK and therefore not available for use in pregnant women. A further novel AF03A (oil in water emulsion) adjuvanted influenza vaccine was given conditional marketing authorisation late in the pandemic (June 2010) for use in Europe, by which time the outbreak was substantially over. The authorisation stated that its use could be considered in pregnancy 
This was an ‘emergency’ pandemic research study, designed and executed in a very short time frame and as such there are a number of limitations. Comparison of cord sample antibody titres with maternal serum antibody titres would have provided additional detail; however, maternal serum was not obtained. Older school age children in the household/or those attending a child care facility and previous maternal seasonal influenza vaccination could have affected the results however, these data were not collected and seasonal influenza vaccine was not recommended in the UK for pregnant women until the 2010–11 winter season, immediately post-pandemic. In order to check for recruitment bias it would have been useful to know how many eligible subjects refused consent or declined involvement; however for logistical reasons this information was not available. Ideally the mothers and children in the study would have been followed up by taking serial serological samples to assess the duration of antibody persistence. This was considered, but dismissed as being impractical. The babies have been followed up separately to assess the clinical protective effect of the maternally transferred immunity. This separate element of the study was designed to obtain nasal mucous samples (for determination of the presence of influenza virus PCR) from babies in the original cohort if a respiratory illness occurred in the follow up period. This secondary objective study recently completed and will be reported separately.
This study was not powered to test for safety issues relating to the novel vaccination using an AS03A-adjuvanted H1N1 2009 vaccine. Nonetheless no serious adverse events were reported to us during the study.
Our study focused on the presence of HI and MN antibody but did not assess clinical effectiveness. The increased rates of illness and death from pandemic influenza during pregnancy make randomised placebo-controlled trials of the efficacy of immunization during pregnancy with H1N1 2009 vaccine unethical, especially as the WHO and national authorities, including the United Kingdom Departments of Health, recommended vaccination during pregnancy as a priority during 2009–10 (pandemic vaccine) and 2010–11 (seasonal trivalent vaccine). An insight into the protection afforded to mothers and their children by vaccination of pregnant women with H1N1 2009 vaccine could perhaps be obtained from effectiveness studies in countries with large databases that capture all relevant information in primary and secondary healthcare settings. Meanwhile, we consider the decision by the Departments of Health in the UK to target pregnant women for vaccination with an AS03A-adjuvanted vaccine was justified from the perspective of the serological protection conferred both to the mothers and their babies. However, we have no equivalent data for the use of non adjuvanted, trivalent seasonal vaccine in pregnancy and so cannot generalize further.
Our work strengthens the need for a better understanding of the relationship between measurable antibody and protective immunity in infancy. Future comparative studies of different vaccines administered during pregnancy to inform vaccine policy are also justified.