Influenza vaccination and children: weighing up the evidence
As has been widely reported, the Food and Drug Administration recently sent CSL a warning letter raising concerns about its manufacturing processes and the company’s investigation into problems with its influenza vaccine for children, Fluvax (which has been marketed in the US as Afluria).
The letter, dated 15 June, asked for CSL to respond within 15 days, so it will be interesting to see if this response is publicly released, or what further material might be released by the FDA.
Meanwhile, epidemiologist Associate Professor Heath Kelly has been investigating how the potential benefits and harms of influenza vaccination of children weigh up, and suggests that the CSL vaccine had specific problems that should not influence considerations about the wider use of influenza vaccination in children.
What is the evidence on influenza vaccination and children?
Heath Kelly writes:
An unexpectedly large number of febrile convulsions (fits due to fever) occurred in young children following immunisation with seasonal influenza vaccine in 2010. Based on a comparison of hospital admissions due to influenza with those due to febrile convulsions caused by the vaccine, this has led to the suggestion that vaccinating healthy children against influenza generally causes more harm than good.
Let’s look at the evidence for this claim.
For children under 5 years of age, this requires us to estimate:
- hospital admissions for influenza, to estimate the burden of influenza
- influenza vaccine effectiveness, to estimate the probability of preventing a hospital admission, and
- the risk of febrile convulsions following influenza vaccination.
Hospital admissions for influenza
In a study from South Australia in children aged 0 to less than 5 the average annual rate of hospital admission due to influenza was about 63 per 100,000 children during 1996-2006, a period when vaccinating children against influenza was rare (http://www.mja.com.au/public/issues/188_09_050508/don11092_fm.html).
The rate of hospital admission for unvaccinated children aged 6 months to less than 5 years in Western Australia during the pandemic of 2009 was 90 per 100,000 (http://www.eurosurveillance.org/ViewArticle.aspx?ArticleId=19661).
In an average influenza season we would expect the hospital admission rate in unvaccinated children to be in the range 50-100/100,000 per year.
It could be much higher in a severe influenza season and lower in a mild season. For instance in WA in 2003, the rate was 205 per 100,000 while it was only 39 per 100,000 the following year.
Influenza vaccine effectiveness
Vaccine effectiveness measures the proportion of infections prevented by vaccination. Two studies published this year provide good evidence for influenza vaccine effectiveness in children.
A study conducted in Western Australia in 2008 in children aged 6 months up to 5 years suggested influenza vaccines prevented about 68% of laboratory confirmed infections in children who had received 2 doses of vaccine.
In Finland, a country that supports universal influenza vaccination for children, influenza vaccine was shown to be 66% effective against symptomatic laboratory confirmed influenza in children aged 9 months up to 4 years, with the same estimate for children younger than 2 years.
Febrile convulsions following influenza vaccine
Although Australia may have sub-optimal methods for monitoring adverse events caused by immunisation (http://immunise.health.gov.au/internet/immunise/publishing.nsf/Content/11DFBB4FD968D072CA25789400172DA1/$File/adverse-event-march-2011.pdf), these monitoring systems are better established in the United States.
The biggest published study from the US found one febrile convulsion in 45,356 children aged 6-23 months vaccinated against influenza. This corresponds to a rate of 2.2 per 100,000 vaccinated children (http://jama.ama-assn.org/content/296/16/1990.short).
The single febrile convulsion in this study occurred on day 3, when it was much less likely to be associated with the vaccine. The febrile convulsions in children following influenza vaccination in 2010 almost all occurred within 12 hours of vaccination.
When assessing the risk of febrile convulsions following influenza vaccine in 2010, the Therapeutics Goods Administration, Australia’s regulatory agency for medicines and therapeutic devices, used unpublished data from the Centres for Disease Control in the US to suggest the expected rate of febrile convulsions on the day of influenza vaccination should be about 3 per 100,000 vaccinated children (http://www.tga.gov.au/safety/alerts-medicine-seasonal-flu-100702.htm).
The risk of febrile convulsion following influenza vaccine over many years in the US was 2-3 per 100,000 vaccinated children.
Putting the numbers together
In a hypothetical population of 100,000 Australian children aged between 6 months (the age at which influenza vaccines are first licensed) and less than 5 years, we might expect around 50-100 hospital admissions during a usual influenza season, based on the risks above.
If we vaccinated all these children with a vaccine with the same safety profile as those used in the US, we might cause 2 or 3 children to have a febrile convulsion. This is calculated from a risk of febrile convulsion of 2-3 per 100,000 in our hypothetical population of 100,000 children. About one third of these children end up in hospital. We may therefore have caused one hospital admission due to vaccination.
Let’s say our vaccine was 60% effective, something less than the two recent studies reported. If we vaccinated all the 100,000 children, we might prevent 30-60 hospital admissions. This is calculated as 60% of the 50-100 unvaccinated children who would otherwise have been admitted to hospital.
We may have caused one admission due to vaccination. For hospitalisation, the benefits would have outweighed the risks by 30-60:1.
In a severe influenza season this ratio would be increased while it would be decreased in a mild season. But it would always be in favour of vaccination if the vaccine had the safety profile from the US studies.
This was not the situation in 2010 in Australia, because the CSL vaccine did not have this safety profile (http://www.eurosurveillance.org/ViewArticle.aspx?ArticleId=19661). Instead of causing febrile convulsions at the rate of 2-3/100,000 expected from the US data, the CSL vaccine caused them at a rate of 440/100,000 (http://bmjopen.bmj.com/content/early/2011/05/28/bmjopen-2010-000016.short).
Why the debate?
There are a couple of reasons for the current debate.
Firstly, we have assumed all influenza vaccines are the same. It turns out that vaccines from different manufacturers are manufactured differently and are not the same. Moreover we now know the CSL vaccines, at least, vary from year to year (http://bmjopen.bmj.com/content/early/2011/05/28/bmjopen-2010-000016.short.)
Secondly, we didn’t know what frequency of febrile convulsions to expect following influenza vaccination of children in Australia. This led to suggestions that the CSL vaccine was associated with a rate of febrile convulsions that was 10 times higher than expected. The actual rate was hundreds of times that expected from US data.
The discussion in 2010 should never have been about immunising pre-school children against influenza. It should have been about using the CSL vaccine to immunise these children. This same vaccine, although still licensed for older children and adults, is no longer recommended for children under 5 years anywhere in the world.
Take home messages
Acceptance is general, but not universal, that influenza vaccines are safe and effective in children. The proviso is that this may not be true for CSL influenza vaccines, as evidenced by recommendations not to use them in young children in Australia, the US and the UK.
We can be sure that influenza causes many hospital admissions in children each year and that immunising children against influenza will prevent more hospital admissions than it causes.
The debate about influenza vaccination in children that started in 2010 is now adversely affecting influenza vaccine uptake in Western Australia (http://www.mja.com.au/public/issues/195_01_040711/letters_bly_fm.html). As happened in New Zealand in 2010, the discussion should have focused on the use of the 2010 CSL vaccine in children, not on the use of all influenza vaccines in children.
Risk-benefit assessments underlying the ongoing debate on the wisdom of vaccinating Australian children against influenza should use data derived from representative influenza vaccine formulations, not the outlier CSL influenza vaccine of 2010.
• Heath Kelly is Head of the Epidemiology Unit at the Victorian Infectious Diseases Reference Laboratory and Associate Professor/Adjunct Associate Professor at School of Population Health, University of Melbourne; School of Computer Science, University of Western Australia; National Centre for Epidemiology & Population Health, Australian National University.
For previous related Croakey posts