Should pregnant mothers be vaccinated against influenza?

This is currently a recommendation in the USA and Canada across all trimesters, but the uptake of this vaccine in the USA is only 16% indicating reluctance amongst pregnant ladies and health professionals. I must admit to being one of the reluctant health professionals and decided to read up on the matter. An article in the Lancet Infectious Diseases reviewing influenza vaccine in pregnancy by Mak et al in January 2008 provided a very good summary of the data concerning this topic so far. It was titled “Influenza vaccination in pregnancy: current evidence and selected national policies” I have summarised the article in a Q & A format with my own comments in italics.

1. How many pregnant women are exposed to influenza?

11%, as worked out by a four fold increase in antibody titres.

2. Does it cause increased mortality in pregnant mothers?

a. When comparing deaths in influenza season (1989-90) to non-influenza season (1985-6) there were 8 pregnant ladies who died in the influenza season compared to 2 in the non-influenza season.

This has been criticised by two other authors Dr Ayoub and Yazbak, who cite a 17 year study examining 38,151 pregnant ladies and not recording a single maternal death. This study itself has been criticised as it relied on patient interviews and recall and was thought to be underpowered. Another criticism of the above data is that the causes of deaths are not recorded though it does provide important information if the deaths are truly attributable to influenza. Further, I wonder whether this trend is held up with other years during the influenza and non-influenza season – no mention of this is made here.

3. What degree of morbidity is associated with influenza?

a. Comparing the peri-influenza season to the influenza season from 1974-93 there was an excess rate of admission due to cardio-respiratory illness in hospitalised pregnant women. The excess rates per 10,000 were x3 x6 x10 for first, second and third trimesters respectively. Compared to non-pregnant women who had a rate of 1.91.

This data was based on Medicaid data, which has a greater representation of lower socio-economic groups and increased smokers. Again the data was examining hospital admission and not identifying influenza per se, but it could be argued that comparing this to a similar time in the year should cancel out the possibility of confounders (i.e. other respiratory infections which also occur at an increased rate during winter seasons).

b. Data from Nova Scotia and a higher socio-economic group did not correlate with the figures given above. The excess rate of hospital admissions comparing a influenza and non-influenza season were 1.1 (-0.1 to 2.3), 0.4 (-1.1 to 1.9) and 2.0 (-0.3 to 4.3) for the three trimesters when comparing pregnant to non-pregnant women.

All the data sets had a range which extended across 0 implying that it was within the range of error rendering these results non-significant. An alternative conclusion is tht women from a high socio-economic background are not disadvantaged by influenza infection during pregnancy.

c. Other data from Washington US HMO databases looked for the incidence of influenza like illnesses and the excess rates of consultations between pregnant and non-pregnant ladies. The excess rates of consultation were: 5.8, 9.8, 14.1 and 11 for the three successive trimesters and post-partum. Only 5.4 % of these consultations were classed as severe.

Consultations marked as influenza-like illness are likely to be a poor surrogate marker for influenza. The data also mirrors the data from Tenesse Medicaid data (3a) and makes you wonder what the socioeconomic group was.

d. Data from Oregon examined the number of outpatient visits for acute respiratory diseases in pregnant v non-pregnant women across four years 1975/6/8/9. In 1978 when a previously non-circulating strain H1N1 resurfaced the excess rate of medical visits by pregnant women was 48. No excess was shown for pregnant women during the other years of the study when H3N2 was predominant.

Again influenza is being bunched together with other respiratory diseases which detracts from this data set. Acute respiratory diseases included all the following: influenza, pneumonia, URTI and respiratory symptoms. Bearing the above point, it seems pregnant ladies during years of pandemics where new strains are circulating may be at more risk than non-pregnant ladies. The converse also holds true, pregnant ladies during normal seasonal influenza are not at increased risk.

4. What are the effects on pregnancy outcomes?

a. From 1998-2002 6,277,508 hospital admissions for pregnant women were examined. Rates of women with influenza or respiratory illness were 2.3% v 1.2% in influenza v non-influenza season. Respiratory disease was associated with the following higher odds (respiratory illness v non-respiratory illness): Preterm delivery 4.08, Fetal Distress 2.48, ‘C’ section 3.91.

This data does not differentiate between pneumonia and influenza. It is more plausible to assume that pneumonia was more strongly associated with adverse peri-natal outcomes.

5. Are women with co-morbidities more at risk?

a. Pregnant ladies with co-morbidities had 3.2 greater rate of respiratory illness than women without co-morbidities during defined influenza months.

Again this study is using respiratory illness rather than influenza.

b. There was 10 fold increased rate of admissions of asthmatic pregnant women v non-asthmatic pregnant women during an influenza season.

Again this looks at a surrogate marker rather than influenza.

c. Novo Scotia 1990-2002 Pregnant ladies with one or more co-morbidities, compared influenza attributable hospital admissions between pregnant mothers with and without influenza. 3.9 (-6.4 to 14), 6.7 (-4.1 to 17) and 35.6 (21 to 50) in the three trimesters.

The results are only significant in the third trimester, and again the influenza attributable

hospital admissions is imprecise.

6. Are women at more risk during pandemics?

a. 1918-19 20 million died. In a mail survey 1350 women were diagnosed with influenza. Fatality rate was 27% but in the pneumonia subgroup it was 54%. Case fatality in non-pregnant ladies was 32% in a series of patients admitted at the same time with pneumonia. In 1957-8 12/103 people who died were pregnant and all had fulminant haemorrhagic pulmonary oedema.

The two patient sets were not age and sex matched. In contrast to 1918 and 1957 in the 1968-9 pandemic there is an absence of evidence of increased risk of morbidity or mortality in pregnant women.

7. Can maternal infection harm the fetus?

a. There are 1 or 2 case reports of in-utero infection of the fetus.

But no IgM Ab in cord sera in 138 infants whose mothers had confirmed  serological infection with influenza. IgM cannot cross the placenta so demonstration of it implies intra-uterine infection. This clearly contradicts the two case reports above.

b. Cluster of 12 deaths in 3 weeks in a family practice. 8 spontaneous abortions and 4 still births, expected rate 12/year. Evidence of serological exposure to Influenza A during pregnancy in all 12 mothers.

But a single cluster is poor evidence as clusters can and do occur due to random chance.

c. 50-100% cases causes a temperature >37.8 lasting 3-5 days with a range of 38-40 C.

No evidence to link pyrexia of the mother with congenital abnormalities.

d. One seroepidemiological study has provided evidence of a higher risk of adult schizophrenia if maternal influenza exposure occurred in the first half of pregnancy.

Schizophrenia has been extensively studied and associated with a multiplicity of factors both genetic and environmental. The multiplicity of associated factors usually through epidemiological studies will forever be plagued by the problem of confounders.  Further given the fact that influenza virus does not cross the placenta(apart from two isolated case reports)  the biological plausibility of such an association  is suspect.

8. What is the risk in the neonatal stage?

a. 5.7% of children in a study examining 2797 children presenting to selected clinics had a positive nasal/throat swab for influenza. Implying it is a significant problem.

b. Lab confirmed hospital admission due to influenza per 1000 are : 0-5m 4.5, 6-23m 0.9, 24-59m 0.3. Implying that there are signifcant rates of morbidity especially in the first 6 months of life enough to cause hospital admission.

c. The rate of influenza in non-hospitalised young children is looked at, children who are 0-5m had the lowest annual rates of outpatient visits with lab confirmed influenza. While those who were 6-23m had the highest rates. There were no differences in these rates inside or outside the influenza season.

The implication from studies (a) and (b) is that neonates will be afforded protection during the first 6 months of life by vaccinating pregnant mothers. But if study (c) is taken into account the event rate, in a group more reflective of the population as a whole, is lower in the age group (0-6 months)  hence the early benefits are mitigated especially as the half-life of IgG is only 45 days.

9a. What is the immunological efficacy of the vaccine in pregnancy?

a. 3-4 weeks post vaccination pregnant and non-pregnant ladies have the same titre of circulating antibodies.

b. IgG is transmitted across the placenta when mother are immunised, with an 87-99% transfer of  IgG occurring. the t1/2 is 43-53 days which is the same as natural antibodies. There is no difference if vaccination occurred in T2 or T3.

9b. What is the clinical efficacy of the vaccine in pregnancy?

a. Inactive vaccine, in non-pregnant recipients, according to Cochrane prevents 67% of serologically confirmed cases and 25% of clinically apparent cases. It is argued that using only recent better studies this figure is higher.

b. A RCT in Bangladesh assessed the prevention of febrile illness if pregnant ladies were immunised in T3, the percentage of febrile illness prevented were 25% (4-46%). Protection of infants was around 61% based on figures from the Bangladesh study.

c. Texas Study looked at children under 6m of age born to immunised mothers. They were less likely to have medically attended respiratory illness during 2004/5 influenza season compared to babies born to non-immunised pregnant women (10.9% v 31% p<0.001)

d. Epidemiological studies followed 39 mother-infant pairs in 1978/9 influenza season. Mothers were infected when pregnant and there was NO reduction in the rate of clinically apparent serologically proven acute infection in infants, but there was evidence to suggest respiratory illness was milder and delayed in onset.

e. Retrospective studies from 1997-2002 in the USA did NOT find a reduction in the incidence of  medically attended respiratory illness in either immunised mother or infants. It is said these studies were underpowered as the overall rate of influenza was lower than predicted.

Study (a) talks about the general efficacy of the vaccine at 67%. The only study which had clearly identified and relevant outcome measure is (b). The one other study (c) in support of clinical efficacy of the vaccine used attended respiratory illness as an outcome measure which is an unclear outcome measure. Study (d) followed pregnant mothers infected by influenza when pregnant, which could be argued as having a highly effective form of immunization and this turned out to be a negative study. Study (e) looking over a much longer period of time are also negative. The evidence to argue for a clear clinical benefit is limited and contradictory.

10. How safe is the vaccine in pregnant women?

  1. 650 mothers were vaccinated in the USA at T1 with Influenza trivalent inactivated vaccine/ IPV/ T Toxoid and diphtheria. They were followed up for 7 years and no association was found with minor or major malformations.
  2. In another study 2991 mothers were vaccinated in T 1/2/3 with a follow up of 1 year for malignancies or increased cancer mortality at 4 years, no significant associations were found.
  3. 1976-77 study looked at 41 mothers vaccinated in T1 and compared them to non-vaccinated mothers. The infants were followed upto 8 weeks for evidence of neurodevelopmental problems and no association was found. A side effect rate of <3% was found to the mothers (fever, headache, and myalgia).
  4. 1998-2003  A Texas study, looked at 252 vaccinated pregnant mothers and compared them non-vaccinated mothers and looked at the hospitalization rate of infants 6 months after their birth. No significant association was found.
  5. 2004-5 A study looked at 1006 vaccinated pregnant mothers and compared them to 1495 unvaccinated pregnant mothers and no serious event in pregnancy was found.
  6. 1976 Study examined 45 pregnant mothers vaccinated in T3, no side effects noted.

The safety studies are clearly positive and encouraging, but are limited by the number of women in each study and the limited duration of follow up or non-specific outcomes in many.

11. Are there any other risks that should be considered?

  1. Fetal hypoxia can occur as a result of anaphylaxis.
  2. The vaccines contain thiomersal (sodium ethylmercurithiosalicylate – a mercury containing compound). It is a component of the inactivated influenza vaccine. Studies in children looking at any causal association with neurodevelopmental disorders have proven to be negative. These have followed children up to 7 years.  The US Institute of Medicine said in 2001 that there is biological plausibility of neurodevelopmental disorders occurring as a result of thiomersal. This decision was reversed in 2004 when they rejected causality as a result of the review of the cumulative paediatric exposure to thiomersal containing vaccines. The European Medical Agency has agreed with them.
  3. The WHO states there is no evidence of the lack of safety of vaccines containing thiomersal but noted that there was a lack of safety data for pre-term and malnourished infants.
  4. The UK health department recommend thiomersal free vaccines for pregnant women but if only thiomersal containing vaccines are available any theoretical
  5. 1994-2004 Yellow Card database: This is a passive reporting system. There are 1366 adverse reports concerning influenza vaccine.  8 of these were in pregnant ladies. 7 had the vaccine in T1. 6 of the mothers had asthma, diabetes or pleurisy. 2 had other vaccines at the same time. 4 had other treatments as well. The adverse outcomes that were reported were as follows: 1 stillbirth, 3 spontaneous abortions, 3 fetal growth retardations of which 2 were premature deliveries, 1 urinary tract anomaly which resolved by birth (thought to be an ultrasound error)

There are no studies looking at the effect of thiomesal per se on human fetal development. Thiomersal free vaccines are offered to women in the UK and are available in the USA. The Yellow card data is definitely concerning and the authors have given it little weight arguing that due the nature of its collection interpretation of the results is difficult. It is worth remembering that the Yellow Card scheme was partly founded to pick up rare but serious adverse associations with drugs such as Thalidomide. Where association is weak but positive the power of a study to pick up such an association may not be possible to perform and therefore any link will technically lie unproven. The Yellow Card data while difficult to interpret cannot be wholly ignored.

12. Which countries or bodies recommend vaccinating pregnant mothers?

  1. WHO : Influenza vaccine in pregnancy is considerd safe and recommended for all pregnant women in the influenza season.
  2. USA: Practiced since the 1950’s. ACIP recommended it in 2004 for all trimesters. Only 16% take up.
  3. Canada: 07-08 vaccine in all trimesters.
  4. Australia: T2/3 recommended
  5. UK: Any trimester if co-morbidities rpesent.
  6. Germany: Does not routinely recommend influenza vaccine. They note that safety evidence is incomplete but no teratogenic effect has been clearly identified.

The German statement on the incomplete data on the safety of the vaccine speaks volumes.

Summary

(1,2) Pregnant women are exposed to the influenza virus, but the case for their increased mortality is based on inadequate data. (3) As for an increased morbidity most of the cited studies are looking at non-specific outcome measures such as consultations, or admissions to hospital for cardio-respiratory disease and are thus limited. Though of note are the increased incidence in lower socio-economic groups and when new strains of the influenza virus are seen circulating. This is most pronounced in the third trimester. (4) While the odds ratios for adverse pregnancy outcomes is significantly positive with respiratory disease inside and outside of the influenza season it is limited again by the non-specific nature of the outcome measures. (5) For women with co-morbidities the rates of admission are more pronounced in the influenza season vs the non-influenza season but suffer from the same imprecision in outcome measures. (6) Pregnant women seem to have a much higher mortality if affected by pneumonia during a pandemic, but again this data is not consistent across al pandemics. (7) There is scant data of any quality to claim an adverse effect on the fetus during maternal infection by the influenza virus. (8) The presumed benefits of maternal vaccination to neonates is potentially present but has to be balanced by the fact that vast majority of infants are not affected by influenza in the primary care setting. (9a) The vaccine is immunologically effective but the evidence of its clinical benefit to the infant post-birth is unclear. (10) The safety studies to date are limited to either 7 years or measuring unclear outcomes, hence the lack of a significant result should be viewed with caution.  (11) The Yellow Card data while difficult to interpret cannot be wholly ignored. Finally the German statement on the incomplete data on the safety of the vaccine speaks volumes.

So what would I do?

I am not convinced about the whole sale recommendation of influenza vaccination of pregnant women primarily due to the incomplete safety data. The benefits of the vaccine are probable but small and the harms while improbable and serious. Weighing the two up the pendulum would swing towards giving it especially in the case of pregnant, from a low socio-economic background with the background of a circulating pandemic with a new strain, in which case I would give the vaccine at the end of T2. And the pendulum swings back to the opposite corner if the pregnant mother is from a relatively high socio-economic background in the absence of a pandemic or a newly circulating strain of the influenza virus. As for the cases in between these two poles the decision is to be made on an individual clinical basis. But to my mind the first Hippocratic principle of ‘Do no harm’ would make the absence of vaccination my preferred choice.

Abbreviations Used

T1 Trimester 1

T2 Trimester 2

T3 Trimester 3

Advertisements

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s