Influenza vaccination: policy versus evidence

Tom Jefferson points out that, for ethical reasons, it is difficult
to conduct a truly random interventional study of influenza vaccine
effectiveness. That is, given current recommendations, it is difficult for
investigators to take a large group of well-matched elderly people, deny
half of them a flu shot, and then wait to see who dies.

Another way to look at his argument is to ask if influenza related
deaths have decreased in age groups most likely to be immunized (persons
over age 65) more than in age groups less likely to be immunized (persons
under age 65). If influenza vaccines are effective, then death rates
should be falling over time in the heavily-vaccinated elderly but
unchanged over time in lightly-vaccinated younger persons – all other
things being equal.

In fact, influenza mortality and the influenza related
hospitalization rates significantly increased for elderly Americans
between 1980 and 2000, a phenomenon only partly explained by the aging
population.(1,2) Increasing hospitalization and death rates among the
elderly stands in stark contrast to annual mortality changes in persons
less likely to be immunized. Excess pneumonia and influenza (P&I) deaths
for persons under age 65 dramatically decreased between 1975 and 1994.(3)
Childhood mortality rates due to all respiratory illnesses fell markedly
in the 1990’s.(4)

If flu shots are effective - and if more older Americans have been
getting them - why have more older Americans been dying of the flu?
Furthermore, why has the P&I death rate in the younger - and less likely
to be immunized - population declined so dramatically during this same
time?

All other things being equal, such changes in P&I death rates are
inconsistent with vaccine effectiveness, indeed they suggest the vaccine
is dangerous. As is often the case in medicine, all other things are not
equal.

Recently, in Epidemiology and Infection, my co-authors and I reviewed
the substantial evidence that suggests vitamin D favorably affect
influenza infection.(5) Indeed, the extant epidemiological and
interventional evidence is so suggestive that we asked, “Is influenza
infection a sign of vitamin D deficiency as much as Pneumocystis carinii
pneumonia is a sign of AIDS.”

As 25(OH)D levels are dependent on surface UVB radiation, increased
surface UVB radiation over time will tend to increase 25(OH)D levels over
time. Surface UVB radiation, as inferred from satellite-based
measurements, has increased since 1990, probably due to ozone
depletion.(6)

Furthermore, the incidence of non-melanoma skin cancer, the skin
cancer most closely associated with sunlight, dramatically increased in
the 1980’s and 1990’s.(7,8) Likewise, cataracts are thought to be related
to UVB radiation. Cataract extraction rates in Minnesota quadrupled
between 1980 and 1992.(9)

Both the increasing incidence of non-melanoma skin cancers and
increasing cataract extraction rates suggest that surface UVB radiation
has significantly increased in the last 20 years. Increasing surface UVB
should also increase average 25(OH)D levels, assuming no change in sun
avoidance or global weather patterns. To my knowledge, no serial
measurements of 25(OH)D levels in stored sera exist to confirm the
hypothesis that increasing surface UVB radiation has translated into
higher year-over-year 25(OH)D levels.

As surface UVB radiation increased, 25(OH)D levels would only have
increased in those whose sun-exposure habits had either increased or not
changed over time. Despite recent government campaigns to limit sun
exposure, the young often ignore that advice while the elderly follow
it.(10) Indeed, one estimate of sun exposure in the young actually
showed it increased during the 1990’s.(11) That is, there are age-
discrepant time trends in sun exposure habits over the last 20 years.

The vitamin D theory of influenza would predict that, despite
influenza vaccines, P&I mortality among the elderly would have serially
increased due to falling 25(OH)D levels, while P&I mortality among the
young would have serially decreased due to increasing 25(OH)D levels. The
age-discrepant time trends in P&I mortality that are being observed are
exactly what the vitamin D theory of influenza would predict.

As Mr. Jefferson reported, "a messy blend of truth conflicts and
conflicts of interest making it difficult to separate factual disputes
from value disputes" in the debate about influenza vaccines. Perhaps the
elderly, virtually all of whom are vitamin D deficient, would be better
off getting an injection of vitamin D in the fall, rather than a flu
shot?(12)

1. Thompson WW, Shay DK, Weintraub E, Brammer L, Bridges CB, et al.
Influenza-associated hospitalizations in the United States. JAMA. 2004 Sep
15;292(11):1333-40.

2. Thompson WW, Shay DK, Weintraub E, Brammer L, Cox N, et al.
Mortality associated with influenza and respiratory syncytial virus in the
United States. JAMA. 2003;289:179-86.

3. Simonsen L, Clarke MJ, Schonberger LB, Arden NH, Cox NJ, et al.
Pandemic versus epidemic influenza mortality: a pattern of changing age
distribution. J Infect Dis. 1998;178:53-60.

4. Panickar JR, Dodd SR, Smyth RL, Couriel JM. Trends in deaths from
respiratory illness in children in England and Wales from 1968 to 2000.
Thorax. 2005;60:1035-8.

5. Cannell JJ, et al. Epidemic influenza and vitamin D. Epidemiol
Infect. 2006 Dec;134(6):1129-40. Epub 2006 Sep 7.

6. Pinker RT, Zhang B, Dutton EG. Do satellites detect trends in
surface solar radiation? Science. 2005;308:850-4.

7. Christenson LJ, Borrowman TA, Vachon CM, Tollefson MM, Otley CC,
et al., Incidence of basal cell and squamous cell carcinomas in a
population younger than 40 years. JAMA. 2005;294:681-90.

8. Karagas MR, Greenberg ER, Spencer SK, Stukel TA, Mott LA.
Increase in incidence rates of basal cell and squamous cell skin cancer in
New Hampshire, USA. New Hampshire Skin Cancer Study Group. Int J Cancer.
1999 May 17;81(4):555-9.

9. Baratz KH, Gray DT, Hodge DO, Butterfield LC, Ilstrup DM.
Cataract extraction rates in Olmsted County, Minnesota, 1980 through 1994.
Arch Ophthalmol. 1997 Nov;115(11):1441-6.

10. Stoebner-Delbarre A, Thezenas S, Kuntz C, Nguyen C, Giordanella
JP, et al. [Sun exposure and sun protection behavior and attitudes among
the French population] Ann Dermatol Venereol. 2005;132:652-7. (abstract
only)

11. Robinson JK, Rigel DS, Amonette RA. Trends in sun exposure
knowledge, attitudes, and behaviors: 1986 to 1996. J Am Acad Dermatol.
1997;37:179-86.

12. Kaplan M. Sunshine may beat the winter flus. Seasonal illnesses
could be down to lack of vitamin D. Nature. 3 November 2006; |
doi:10.1038/news061030-12

Competing interests:
Director, The Vitamin D Council

Jefferson’s article attempts to use evidence, ours included, to
question the seasonal influenza vaccination campaign, but raises a number
of issues. Firstly the article is not one that follows established
Cochrane procedures in that it provides incomplete evidence. Although
noted in table 2, Thomas Jefferson does not discuss the high, over 70%,
efficacy against laboratory confirmed illness from influenza by the
trivalent inactivated vaccine in healthy adults and children in the
Cochrane reviews or, in fact, other meta-analyses (e.g. reference
1appendix 20 page 249). In a trial in over 60 year olds vaccine efficacy
was only slightly lower at 58% (95% CI 95% CI 26-77%))2. This last trial
is not mentioned in the article but is key. As there is clear evidence
of benefit against laboratory confirmed illness from influenza vaccination
including in over 60 year olds, a trial now, with mortality as an
endpoint, would not be ethical. Instead we have to rely on observational
studies - which Thomas Jefferson dismisses. Such studies have an
important role but we have to deal with the following constraints: the
presence of both positive and negative confounding and the restriction to
easily measured outcomes.

It is well known that vaccine recipients may be different to non-
recipients in many ways. In some populations there may be a “healthy
vaccine effect” with non-smokers and those with higher educational or
social status more likely to have influenza vaccine3;4. This positive
confounding would bias upwards any protective effect of the vaccine.
Negative confounding “by indication”, in which those identified as more
frail are more likely to be offered the vaccine, would bias downwards any
protective effect of the vaccine. The predominance of negative
confounding explains some of the crude estimates of effect showing no
effect of the influenza vaccine in the table in the Cochrane review of
influenza vaccine in the elderly5.

Statistical methods are commonly used to control for both positive
and negative confounding. By doing so a protective effect obscured by
negative confounding is then possible to see. Residual confounding is
however often a problem because measured factors are not able to capture
all the differences between vaccine recipients and non recipients.
Residual negative confounding will act to underestimate an effect. Less
well recognised though, is the scope to over-estimate the true benefit of
an intervention if there is residual confounding by the “healthy vaccine”
effect. Luckily the presence of residual confounding can be assessed by
seeing if there is any effect of the vaccine in seasons where no influenza
is circulating. This method was first used for influenza vaccine by Ohmit
and Monto looking at hospital admissions in the elderly. A 31% protective
effect against hospitalisation for pneumonia and influenza was noted that
was not seen in the corresponding peri-influenza winter season6. More
recently we conducted a large UK cohort study using the General Practice
research Database over several years that included over 2 million person-
years of follow-up. Overall vaccine effectiveness was 21% (95% CI 17-26%)
for respiratory disease hospitalisations and 12% for respiratory deaths
(95% CI 8-16%) in over 64 year olds, with no protective effect seen in the
corresponding non-influenza winter season. In contrast a protective
effect against all-cause mortality was seen in the non-influenza season
suggesting a “healthy vaccinee” effect for that outcome. This protection
against respiratory disease deaths, after adjustment for confounding, is
missing from the full Cochrane review7 and incorrectly noted as not
significant in the summary paper5. An analysis with similar logic found
that though unvaccinated elderly persons showed mortality peaks following
peaks of influenza circulating in the community, those vaccinated were
substantially protected from these mortality peaks. 8 It would have been
preferable that the systematic review of observational studies in the
elderly that Thomas Jefferson refers to had conducted a more rigorous
assessment of the methods studies used to deal with confounding. As
experts in systematic reviews have pointed out, the presence of
heterogeneity in the results of studies, especially observational studies,
should be carefully examined rather than dismissed9 .

The final constraint of observational studies also applies to trials
without laboratory confirmation of the aetiological agent -
misclassification error in measuring outcome. Such error can be easily
shown to under-estimate any effect of an exposure 10. Given outcomes that
are inevitably non-specific it is thus not surprising that estimated
effects of vaccine are often low. A related issue is Jefferson’s
inappropriately dismissive interpretation of the modest effect of the
influenza vaccine. A just over 20% protective effect on a non-specific
outcome such as an admission for acute respiratory disease must reflect a
much larger effect in more specific outcomes.

Finally the article suggests that influenza vaccination requires
resources which could be used for other interventions. Seasonal influenza
can be mild. It is certainly so in some years but not in others. The
seasonal epidemic in 1989/90 is but one example with about 18,000 excess
deaths in the UK after taking into account a slight deficit of deaths
after the epidemic11 ; with an average of over 12,000 deaths per year
when crudely compared with death rates in similar weeks in other years
when influenza is not circulating12. These are rough estimates as
laboratory tests to confirm seasonal influenza as the cause of death are
not usually done. Because the severity of seasonal influenza as a result
of antigenic drift cannot be easily predicted, yearly influenza
vaccination is required. Cost-effectiveness studies of influenza vaccine
have been conducted with sensitivity analyses of the results to varying
the attack rate or assuming no deaths occurred1. In those at high risk
for complications of influenza, including over 64 year olds, the cost per
quality adjusted years of life saved of vaccination is under a few
thousand pounds, well below the accepted threshold for funding in the
heath sector in the UK including smoking cessation activities or breast
cancer screening13.

In short, inactivated trivalent influenza vaccines are highly
effective against laboratory confirmed influenza, with more evidence in
younger adults but also clear evidence in elderly people. In addition
observational studies have shown that the influenza vaccine currently in
use prevents not only hospitalisations but also death in over 64 year
olds. It is not clear why the BMJ should publish such a flawed article,
contrary to the judgement of virtually all other scientists who have
looked at the question, particularly at a time of year when optimising
influenza vaccine coverage can save lives.

Reference List

(1) Turner D, Wailoo A, Nicholson K, Cooper N, Sutton A, Abrams K.
Systematic review and economic decision modelling for the prevention and
treatment of influenza A and B. Health Technol Assess 2003; 7(35):iii-
xiii, 1.

(2) Govaert TME, Thijs CTMCN, Masurel N, Sprenger N, Dinant GJ. The
efficacy of influenza vaccination in elderly individuals: a randomised
double-blind placebo-controlled trial. JAMA 1994; 272:1661-1665.

(3) Christensen B, Lundbergh P. Comparison between cohorts
vaccinated and unvaccinated against influenza and pneumococcal infection.
Epidemiol Infect 2002; 129:515-524.

(4) Mangtani P, Breeze E, Kovats S, Ng ES, Roberts JA, Fletcher A.
Inequalities in influenza vaccine uptake among people aged over 74 years
in Britain. Prev Med 2005; 41(2):545-553.

(5) Jefferson T, Rivetti D, Rivetti A, Rudin M, Di PC, Demicheli V.
Efficacy and effectiveness of influenza vaccines in elderly people: a
systematic review. Lancet 2005; 366(9492):1165-1174.

(6) Ohmit SE, Monto AS. Influenza vaccine effectiveness in
preventing hospitalization among the elderly during influenza type A and
type B seasons. Int J Epidemiol 1995; 24:1240-1248.

(7) Rivetti D, Jefferson T, Thomas R, Rudin M, Rivetti A, Di PC et
al. Vaccines for preventing influenza in the elderly. Cochrane Database
Syst Rev 2006; 3:CD004876.

(8) Armstrong B, Mangtani P, Fletcher AE, et al. Influenza
vaccination protects against excess deaths occurring during periods of
high circulation of influenza. BMJ 2004; 329:660-661.

(9) Egger M, Smith GD, Altman D. Systematic reviews in health care:
meta-analysis in context. 2nd Ed ed. London: BMJ; 2001.

(10) Mertens TE. Estimating the effects of misclassification. Lancet
1993; 342:418-421.

(11) Ashley J, Smith T, Dunnell K. Deaths in Great Britain
associated with the influenza epidemic of 1989/90. Population Trends 1991;
65:16-20.

(12) Fleming DM. The contribution of influenza to combined acute
respiratory infections, hospital admissions and deaths in winter.
Communicable Disease and Public Health 2000; 3:32-38.

(13) Mason J, Drummond M, Torrance G. Some guidelines on the use of
cost effectiveness league tables. BMJ 1993; 306:570-572.

Competing interests:
None declared

When you own a hammer, everything in the world resembles a nail. When
you are an expert in the analysis of double-blind randomized placebo
controlled trials, it must be frustrating to find an area with limited
contemporary data; such frustration is evident throughout Dr. Jefferson’s
recent article on influenza vaccine.

Vaccines are evaluated for efficacy and safety by means of double-
blind randomized placebo controlled trials; witness the recent results
from the novel vaccines against human papillomaviruses (1,2). Yet, once
the vaccine has been licensed, there is an ethical limit to the additional
studies which can be placebo controlled. The influenza field is
particularly difficult because, as Dr. Jefferson summarizes, strains of
influenza virus change annually, as does the vaccine and there are varying
degrees of mismatch between the vaccine deployed and the vaccine which
actually circulates subsequently in any one season. There is also the
problem of the non-specific clinical diagnosis of influenza or influenza-
like illness.

We would all like to have more data in this area, but one statement
in the article concerns me: “A meta-analysis of inactivated vaccines in
elderly people showed a gradient from no effect against influenza or
influenza-like illness to a large effect (up to 60%) in preventing all-
cause mortality. These findings are both counterintuitive and implausible,
as other causes of death are far more prevalent in elderly people even in
the winter months. It is impossible for a vaccine that does not prevent
influenza to prevent its complications, including admission to hospital.”
In contrast to what Dr. Jefferson says, it is well recognized that viruses
can trigger adverse outcomes even if they do not declare themselves
clinically with their classical syndromes. The resulting indirect effects
can nevertheless be prevented by Dr. Jefferson’s beloved double-blind
randomized placebo controlled trials using drugs active against the
provoking viruses (reviewed in 3). Furthermore, additional studies show
that the inflammatory response to respiratory tract infection, but not to
vaccination, are associated statistically with an increased risk of
myocardial infarction and stroke (4) while epidemics of proven influenza
infection are associated with episodes of death attributed to ischaemic
heart disease (5). All of this information provides a plausible
explanation for the reported survival benefit associated with receipt of
influenza vaccine.

1. Harper DM, Franco EL, Wheeler C, Ferris DG, Jenkins D, Schuind A,
et al. Efficacy of bivalent L1 virus-like particle vaccine in prevention
of infection with human papillomavirus types 16 and 18 in young women: a
randomised controlled trial. Lancet 2004; 364(9447): 1757-65.

2. Villa LL, Costa RL, Petta CA, Andrade RP, Ault KA, Giuliano AR,
Wheeler CM, et al. Prophylactic quadrivalent human papillomavirus (types
6, 11, 16 and 18) L1 virus-like particle vaccine in young women: a
randomised double-blind placebo-controlled multicentre phase II efficacy
trial. Lancet Oncology 2005; 6(5): 271-8.

3. Griffiths PD. The indirect effects of virus infections. Rev Med
Virol 2003; 13: 1-3.

4. Smeeth L, Thomas SL, Hall AJ, Hubbard R, Farrington P, Vallance P.
Risk of myocardial infarction and stroke after acute infection or
vaccination. N Engl J Med 2004; 351(25): 2611-8.

5. Fleming DM, Cross KW, Pannell RS. Influenza and its relationship
to circulatory disorders. Epidemiology Infection 2005; 133(2): 255-62.

Competing interests:
I receive influenza vaccine annually. I am a member of the Department of Health Joint Committee on Vaccination and Immunisation, although I am writing this reply in a personal capacity.

G H Hall complains about secrecy regarding UK flu and flu vaccination
data. Some information is available from Government websites, but it is
anomalous and confusing to say the least.

According to the Chief Medical Officer's publication 'Explaining
Pandemic Flu'(1 March 2005):

"Ordinary flu occurs every year during the winter months in the UK.
It affects 10-15% of the UK population, causing around 12,000 deaths every
year." [1]

On the other hand according to another official publication 'Summary
of flu immunisation policy' (3 October 2004):

"Even during a winter where the incidence of flu is low, 3-4000
deaths may be attributed to 'flu; this can rise much higher in epidemic
years, for example there were an estimated 13,000 deaths in 1993 which
were attributable to 'flu and 29,000 in 1989/90." [2]

Thus according to first statement flu causes 12,000 deaths every
year, while according to the second statement in low years the figure is 3
-4,000 but unlike the 12,000 this is not even a hard figure, only "may
be".

We need real figures, and clear criteria. It is also surely essential
to know what impact the vaccination campaign has on mortality. The BBC
reported that take up in the target vulnerable population in 2004-5 was
71.5% [3]. Did this have any impact on the alleged annual death-toll of
12,000? Can it be shown that flu mortality was substantially less in the
vaccinated amongst the target population, as opposed to the unvaccinated,
and what was the effect on general mortality?

If the Government cannot provide transparent data - only fragments
without context - the question remains "why".

[1]
http://www.dh.gov.uk/PublicationsAndStatistics/Publications/

PublicationsPolicyAndGuidance/PublicationsPAmpGBrowsableDocu

ment/fs/en?CONTENT_ID=4106931&MULTIPAGE_ID=5093031&chk=2UuNhx

[2]
http://www.dh.gov.uk/PolicyAndGuidance/HealthAndSocialCareTo

pics/Flu/FluGeneralInformation/FluGeneral/fs/en?

CONTENT_ID=4001688&chk=BbJebs

[3]'Q & A: Flu Jabs', 23 November 2005:
http://news.bbc.co.uk/1/hi/health/4459140.stm

Competing interests:
None declared