Does the Flu Vaccine Increase Hospitalizations?

The Centers for Disease Control (CDC) currently recommends that children 6 month to 18 years old receive an annual flu vaccine. There are two types of flu vaccines used in the US: a live attenuated virus (LAIV) and a trivalent inactivated virus (TIV) vaccine. Both are safe and effective  – while efficacy varies from year to year, they are 70-90% effective in healthy adults. Efficacy is young children appears to be slightly less, about 66%.

There remains, however, many sub-questions about the flu vaccines and by the time researchers have thoroughly explored them vaccine technology is likely to have progressed, and therefore any new vaccines will have to be tested all over again.

One of those sub-questions about vaccine safety and efficacy is the net effect of the flu vaccine in children with asthma.  Some have raised concerns that the vaccine may exacerbate asthma, a 1-2% increased wheezing and 3% increased hospitalizations have been reported, although so far the bulk of the data suggests that both types of flu vaccines are safe in children with asthma.  There is evidence to suggest that the LAIV may be superior to the TIV in children, particularly with asthma.

A new study, presented but not published, further explores the safety and efficacy of the TIV in children.  Study author, Avni Joshi, M.D., of the Mayo Clinic, reports:

“The concerns that vaccination maybe associated with asthma exacerbations have been disproved with multiple studies in the past, but the vaccine’s effectiveness has not been well-established.  This study was aimed at evaluating the effectiveness of the TIV in children overall, as well as the children with asthma, to prevent influenza-related hospitalization.”

The study is a retrospective study of 263 children who presented to the Mayo clinic with laboratory confirmed influenza. They found that children who had recieved the TIV vaccine had a 3 times greater risk of hospitalization than those who were not vaccinated. These results raise concerns about the safety and effectiveness of the TIV in children with asthma.

Dr. Joshi concludes:

“While these findings do raise questions about the efficacy of the vaccine, they do not in fact implicate it as a cause of hospitalizations.  More studies are needed to assess not only the immunogenicity, but also the efficacy of different influenza vaccines in asthmatic subjects.”

That may seem like a curious conclusion given the results of this study, but it is accurate. The key to understanding the implications of this study is that it is retrospective. That means it looks at children who have the flu and then looks back to see who was vaccinated and who wasn’t. This in turn means that children were not randomized to either be vaccinated or not, and this opens the door to any number of variables that cannot be controlled for in the study.

The authors did look as obvious factors, such as severity of asthma and insurance status, and found that they did not correlate with risk of being hospitalized. But what other factors might there be? The flu vaccine is optional, which means that parents decide whether or not to vaccinate their children, perhaps with advice from their pediatrician. It is likely that sicker or more frail children are more likely to get vaccinated. It is also likely that children who had a bad reaction to the flu in the past are more likely to get vaccinated. The flu vaccine is recommended especially for those who are at high risk for complications if they get the flu.

Therefore while this study raises important questions, it is not designed to answer them definitively. A prospective trial is required for that, and that is what Joshi means by “more studies are needed.”  In general, retrospective studies are useful to find correlations and generate hypothesis, but are not capable of determining causation – there are simply too many variables that are not controlled for.

As expected, the anti-vaccinationists have already jumped on this study and misinterpreted its significance.  They did not recognize its retrospective nature nor put it into the context of existing research on the safety and efficacy of the flu vaccines.

Clinical trials are complex, and there are many types that each have their own strengths and weaknesses. Often, many independent lines of basic science and clinical evidence need to be brought together to form a reliable conclusion about a specific intervention. That is the essence of science-based medicine. Individual studies typically only provide a tiny slice of information, but are often presented to the public as if they are definitive. This creates a constant background noise of misinformation about medical questions.

It also provides a rich source of data from which to cherry pick, allowing proponents to support almost any notion by shopping from the vast store of often conflicting medical research. This reinforces the need to look thoroughly at the totality of scientific evidence on any claim or question.

When that is done on the question of the flu vaccines, it is clear that both types of vaccines are safe and effective. However, there is also much room for improvement in the vaccine technology itself, as well as evidence-based recommendations for who, exactly, should get which type of vaccine.

This current study adds incrementally to our knowledge on this question, and suggests questions for future research. It is not the kind of evidence, however, that should lead to changes in the current recommendations.

Posted in: Vaccines

Leave a Comment (21) ↓

21 thoughts on “Does the Flu Vaccine Increase Hospitalizations?

  1. ghrasko says:

    Uhhh, if I understand correctly, the original article in ScienceDaily is VERY misleading. It says that “children who had received the flu vaccine had three times the risk of hospitalization, as compared to children who had not received the vaccine” . For an ordinary person it means “STAY AWAY OF VACCINES”!

    On the other hand you say the vaccine’s efficacy is 66% that means – what in fact? Of 100 non-vaccinated X will receive flue, and of 100 vaccinated (1-0.66) X will receive the flue – am I correct?

    Your evaluation of the problem with the study (or in fact with its interpretation) is clear – I just wonder how could they create such a mess?

  2. superdave says:

    This is such a perfect example of why we need good science journalism. The implication that the vaccines caused the flu here seems so obvious and the reason this is false is subtle. Research presented at conferences and in Journals are not designed for the layperson in mind. While there is nothing wrong with the layperson trying to read them, they must take into consideration that they are not the intended audience and may not have the right background to fully interpret the results.

  3. Harriet Hall says:

    This seems to conflict with a large body of evidence showing that people who have been immunized and then get the disease usually have a milder course.

    I’m puzzled that they seem to be defining “effectiveness” as preventing hospitalization; I thought it meant preventing the disease.

  4. “I’m puzzled that they seem to be defining ‘effectiveness’ as preventing hospitalization; I thought it meant preventing the disease.”

    Actually, you can’t tell from the article in ScienceDaily how they defined effectiveness, how they arrived at their conclusions, how they picked the case controls, or several other key things. Those’ll just have to wait for publication, it seems.

  5. Right – it’s always difficult interpreting a study that is not published with full details. But since it’s in the news and the anti-vax sites, it’s worth putting into perspective.

    For me the big issue is that we often will only vaccinate those who are high risk for having a bad outcome if they get the flu. So those who get the flu despite being vaccinated are already pre-selected for being high risk.

    I wonder if/how they controlled for this.

  6. David Gorski says:

    When in doubt, look up the abstract. I checked out the American Thoracic Society website. Here’s the abstract:

    [3:20 pm] Flu Vaccination in Asthmatics: Does It Work?, [Publication Page: A5118]

    A.Y. Joshi, MD, V.N. Iyer, MD,MPH, M.F. Hartz, MD, G.W. Volcheck, MD,Ph.D, A.M. Patel, MD, J.T. Li, MD,Ph.D, Rochester, MN

    INTRODUCTION: Influenza is known to be associated with asthma exacerbation but the effectiveness of the trivalent inactivated flu vaccine (TIV) in asthmatics is unknown.

    METHODS: We conducted a cohort study of all pediatric subjects( 6 months to 18 years age) who were evaluated at Mayo Clinic, Rochester, MN, USA who had laboratory confirmed influenza during each flu season from 1999-2006 to evaluate the efficacy of TIV. A case control analysis was performed with the cases and the controls being the subjects with asthma who did and did not required hospitalization with the influenza illness respectively.

    RESULTS: There were 236 subjects with laboratory confirmed influenza from 1996-2006.

    In assessing the effectiveness of the TIV for preventing hospitalization with influenza in all subjects, there was an overall trend towards higher rates of hospitalization in subjects who got the TIV as compared to the ones who did not get the TIV( OR:2.97, CI: 1.3,6.7).Using Cochran-Mantel-Haenszel (CMH) test for Asthma status stratification, there was a significant association between hospitalization in asthmatic subjects and TIV (P=0.006).
    In the asthmatic subset:

    There was no association between ER visit and receiving the TIV ,severity of asthma and the risk of hospitalization or the hospital length of stay and receiving the TIV.
    In assessing access to medical care, there was no association between hospitalizations and health care insurance plans (Odds ratio:0.3, P= 0.13)

    1) TIV did not provide any protection against hospitalization in pediatric subjects’ esp. children with asthma. On the contrary, we found a 3- fold increased risk of hospitalization in subjects who did get the TIV vaccine.This may be a reflection not only of the vaccine effectiveness but also the population of children who are more likely to get the vaccine.
    2) More studies are needed to assess not only the immunogenicity but also efficacy of different influenza vaccines in asthmatic subjects.

    Am J Respir Crit Care Med 179;2009:A5118

    Day/Date: Tuesday, May 19, 2009
    Session Time: 1:30 PM – 4:00 PM
    Presentation Time: 3:20 PM
    Room: Room 3 (Upper Level), SAN DIEGO CONVENTION CENTER

    Can you all see the problem with the study?

  7. Dr Benway says:

    I dunno. But the design isn’t what I expected: “A case control analysis was performed with the cases and the controls being the subjects with asthma who did and did not required hospitalization with the influenza illness respectively.”

    Most case-control studies separate the cases and the controls by the independent variable (vaccination), then compare measures of the dependent variable (hospitalization for asthma attacks).

    So I wonder, was hospitalization actually the independent variable prior to the various statistical tests?

    Multiple post-hoc analyses of a data set usually turn up something significant.

    I take it that all cases and controls got sick with the flu, so effectiveness of the vaccine wasn’t evaluated.

  8. “This may be a reflection not only of the vaccine effectiveness but also the population of children who are more likely to get the vaccine.”

    Right – that’s how I read it. The result can be entirely due to the population of children more likely to get the vaccine, so I don’t think any other conclusions can be derived from this.

    This is actually not even the kind of analysis that would answer the first question – vaccine effectiveness.

    Thanks for the abstract, David.

  9. David Gorski says:

    Yeah, this study is one big exercise in selection bias.

  10. daedalus2u says:

    A possible confounder is using only laboratory confirmed cases of flu. If someone had a very mild case, there would be no reason to do the test for a differential diagnosis. If someone has a case of some respiratory viral disease that is getting better there is no reason to test to see what it is. Mild cases are also more likely to be false negatives.

    If they only had 236 cases of confirmed flu over a 10 year period that is a pretty sparse data set. If there was no correlation with severity and vaccination status, the association with hospitalization is likely an artifact, not that vaccination made people either more susceptible or sicker.

    Since the flu vaccine is new every year (because flu strains change), it is really looking at 10 different vaccines. If any of those individuals are repeats, that may indicate an idiosyncratic immune response in that individual, not a generic effect of the vaccine.

  11. “There was no association between…severity of asthma and the risk of hospitalization…”

    If I read the abstract correctly (it leaves room for ambiguity), that passage means that the study did NOT find that the increased rate of hospitalization in TIV subjects could be explained by exacerbations of asthma—contrary to the premise in the introduction.

  12. Oops, I read it wrong…

  13. …the premise in the introduction is that INFLUENZA, not TIV, is associated with an exacerbation of asthma. Nevertheless, this study seems not to support others’ “concerns that the vaccine may exacerbate asthma” [in those subjects who were hospitalized, anyway], as mentioned by Steve in his post.

  14. The Blind Watchmaker says:

    In our practice, it seems that the parents of sicker children are more likely to accept the flu shot for their kids than the parents of healthier kids, or the mild asthmatics. Given this, it seems fairly predictable that kids who had the shot are more likely to need hospitalization when they do get the flu. This cohort would likely represent the 35% or so (within the subgroup of “sicker” kids) for which that the flu shot is ineffective.

    At any one time, the population inside a hospital represents the sickest people. These are the one’s most likely to have been vaccinated.

    This study really shows nothing other than what one would expect given the premises.

    We need a prospective randomized controlled trial to conclude anything about the risk of hospitalization, severity of asthma attacks and the flu shot. My money is on the square that says that the net risk is actually decreased.

  15. Wholly Father says:

    There is potential bias in utilization of resources:

    Parents who do not vaccinate may, for a variety of reasons, be more likely to seek care in an ER for less severe illness, therefore, a lower proportion of their ER visits result in hospital admission.

    Parents who vaccinate may be more likely to reserve ER visits for true emergencies, thus a higher proportion or ER visits result in hospital admissions.

  16. Th1Th2 says:

    Yes it does because the flu vaccine definitely can cause the disease itself. This is a no-brainer.

  17. Chris says:


    Yes it does because the flu vaccine definitely can cause the disease itself. This is a no-brainer.

    Do you have any evidence for that statement? If not, I would bet that thinking a vaccine causes the actual disease is can only be concluded by someone without a brain.

  18. Th1Th2 says:

    I would suggest that you get hold of any flu vaccine package insert and read about all the INTENDED side effects for that matter, for vaccination is the introduction of pathologic evidence of the disease. What goes in, goes out.

  19. becunnin says:

    The problem that I see with the study (unless I missed something) is that they did not reveal if they tested to see which strain of flu the hospitalized asthmatic children were infected with and if that strain matched the strains covered in the flu vaccine.

    Unless that variable is distinguished amongst the hospitalized patients, I don’t see how the report can reveal anything substantial about the flu vaccine.

    If the several of those children were infected by an influenza strain not covered in the vaccine or one that was a variant of the strain in the vaccine-then they would not be as protected as if the vaccine was a match for the strain they were infected with.

    Another variable that I think should be considered is the latest studies revealing that the live-attenuated virus offers more immunity than TIV. The issue could be that TIV is not as effective as the live-attenuated version but does offer some immunity.

Comments are closed.