BudWiederman, MD, MA, Evidence eMended Editor, Grand Rounds
Most healthcare providers are aware of the lack of effectiveness of the live attenuated
influenza vaccine (LAIV), apparent in recent flu seasons. Some also were surprised
by what appeared to be a relatively low effectiveness rate for the other flu vaccines.
I'll try to explain all this here, using this recently published, important study.
The current study is a prospective observational study enrolling subjects 6 months
of age and older (including adults) who presented to designated study ambulatory care
clinics for acute respiratory illness during the 2015-16 flu season. Of the almost
7000 participants, a little over 1300 were positive for influenza by PCR testing and
comprised the case group. They were compared to the control subjects who tested negative.
Key findings of the study were that overall effectiveness of all vaccines was 48%
(95% confidence interval 41-55), but LAIV had no effectiveness (5%; 95% CI -47 to
39) for children 2 - 17 years of age. That's why, together with other studies showing
similar results, LAIV is not recommended for use in the United States.
The first issue in attempting to understand vaccine effectiveness (VE) studies is
to look at study design types. Probably the most accurate design to establish VE is
the randomized controlled trial (RCT). RCTs are expensive, and flu poses additional
problems when considering RCTs. First, it would be unethical to perform such a study
with a placebo group, since the evidence of flu vaccine benefits overall is very solid.
Comparing 2 (or more) different flu vaccines in an RCT is fine, but the number of
participants needed for adequate study power (to detect a difference if such exists)
would be enormous, and therefore too expensive. So, the next best design type is the
case-control study. As an example, this could involve finding cases of influenza,
then matching each case to 1 or more controls, without influenza, and comparing vaccine
status. The problem with this approach is that it likely introduces confounding variables
that can't be measured easily. A big issue is that patients who have received influenza
vaccines have different behaviors in their approach to accessing healthcare than do
individuals who do not choose vaccination; they can be more or less likely to seek
medical attention for a flu-like illness. It's hard to quantify the degree of confounding,
adding bias to the study results.
Enter the test-negative design (TND). Some researchers have applied the term test-negative
case-control design to this entity, but that's not quite correct. A case-control study
first identifies cases, but with a TND study the case/control status (positive influenza
PCR, for example) is determined after study enrollment. That is thought to be a major
strength of TND since it is less likely to be subject to bias from differences in
care-seeking behavior. Still, this is a relatively new approach to determine VE, and
thus we shouldn't be surprised if better methodology emerges in the next decade or
Let's get back to the key findings of this study. First, LAIV performed terribly,
especially in children (note that LAIV is not approved above 49 years of age). Although
the biologic reason for this failure isn't certain, it doesn't appear to be due to
a mismatch of the vaccine strains with the circulating flu strains for that year.
That leaves the next most likely reason as "poor replicative fitness," meaning that
the vaccine strains don't replicate well in the human host. (Live viral vaccines need
to multiply in the recipient; that's how immunity is produced.)
The second finding, alarming to some frontline providers, is the relatively low rate
of flu VE overall: 48%. That's much lower than our other vaccines, where we routinely
expect rates above 90% for the regimens we use. Why do accept such low VE in a universally
recommended vaccine? Part of the reason lies in what outcome is being measured. Note
that here, and in many studies of flu vaccine, the outcome is medically-attended influenza,
i.e. a respiratory illness that is severe enough for the patient to seek medical attention.
That's different from an outcome of infection alone, where vaccine recipients (and
controls) are tested at regular intervals regardless of symptoms, and would, of course,
show lower VE. A large benefit of flu vaccination is protection against more serious
illness resulting in hospitalization or mortality, predominantly occurring in very
high-risk individuals including responders to immunization due to altered immune status.
Universal immunization helps protect those folks via herd immunity. So, while it would
be great to have flu VE above 90% for medically-attending illness, a lower but still
significant VE is a great return on investment.
With the birth of my first grandchild approaching soon, I now have an additional reason
to be immunized for flu, not that I needed one. Please be sure to get your flu vaccine
now, if you haven't already.