Date post: | 05-Apr-2018 |
Category: |
Documents |
Upload: | mihail-raicis |
View: | 218 times |
Download: | 0 times |
of 74
7/31/2019 Demicheli Et Al_2000_Prevention and Early Treatment of Influenza in Healthy Adults
1/74
7/31/2019 Demicheli Et Al_2000_Prevention and Early Treatment of Influenza in Healthy Adults
2/74
Contents
1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 959
1.1. Prevention and early treatment of inuenza. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 959
1.1.1. Inuenza vaccines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 959
1.1.2. Ion channel inhibitor antivirals (Amantadine and Rimantadine) . . . . . . . . . . 961
1.1.3. Neuraminidase inhibitor antivirals (NIs) . . . . . . . . . . . . . . . . . . . . . . . . . . . 962
1.2. Rationale for the economic evaluation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 962
2. Methods. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9632.1. Methods for the reviews. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 963
2.1.1. Objectives. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 963
2.1.2. Selection criteria . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 963
2.1.3. Search strategy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 963
2.1.4. Methods for the economic evaluation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 966
3. Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 967
3.1. Results of the reviews . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 967
3.1.1. Description of studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 967
3.1.2. Methodological quality of included studies . . . . . . . . . . . . . . . . . . . . . . . . . 970
3.2. Eects of inuenza vaccines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 971
3.2.1. Eect of vaccination on clinical cases of inuenza . . . . . . . . . . . . . . . . . . . . 971
3.2.2. Eect of vaccination on serologically conrmed cases of inuenza . . . . . . . . 971
3.2.3. Eect of vaccination on other outcomes . . . . . . . . . . . . . . . . . . . . . . . . . . . 971
3.2.4. Recommended vaccines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 972
3.2.5. Vaccine matching the circulating strain . . . . . . . . . . . . . . . . . . . . . . . . . . . . 972
3.3. Eects of amantadine and rimantadine . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 972
3.3.1. Comparison A oral amantadine compared to placebo in inuenza
prevention . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 974
3.3.2. Comparison B oral rimantadine compared to placebo in inuenza
prevention . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 974
3.3.3. Comparison C oral amantadine compared to oral rimantadine in inuenza
prevention . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 975
3.3.4. Comparison D oral amantadine compared to placebo in inuenza
treatment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 975
3.3.5. Comparison E oral rimantadine compared to placebo in inuenza
treatment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9763.3.6. Comparison F oral amantadine compared to oral rimantadine in inuenza
treatment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 976
3.3.7. Comparison G oral amantadine compared to oral aspirin in inuenza
treatment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 977
3.3.8. Comparison H inhaled amantadine compared to placebo in inuenza
treatment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 977
3.4. Eects of NIs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 977
3.5. Results of the economic evaluation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 979
4. Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 979
4.1. Inuenza vaccines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 979
4.2. Amantadine and rimantadine . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 981
4.3. Neuraminidase inhibitors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 982
4.4. Overall comments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9824.5. Economic evaluation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 982
Acknowledgements. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 983
Appendix A. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 984
Appendix B . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 985
Appendix C. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 986
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1030
V. Demicheli et al. / Vaccine 18 (2000) 9571030958
7/31/2019 Demicheli Et Al_2000_Prevention and Early Treatment of Influenza in Healthy Adults
3/74
1. Introduction
Inuenza is an acute respiratory infection caused
by a virus, of which three serotypes are known (A, B
and C). Inuenza causes an acute febrile illness with
myalgia, headache and cough. Although the median
duration of the acute illness is three days (duration
can vary between serotypes and subtypes), cough andmalaise can persist for weeks. Complications of inu-
enza include otitis media, pneumonia, secondary bac-
terial pneumonia, exacerbations of chronic respiratory
disease and bronchiolitis in children. Additionally,
inuenza can cause a range of non-respiratory com-
plications including febrile convulsions, Reye's syn-
drome and myocarditis [1].
The inuenza virus is composed of a lipid mem-
brane surrounding a protein shell and a core consist-
ing of several RNA complexes. On the lipid
membrane are two viral glycoproteins which act as
powerful antigens: neuraminidase (N antigen) andhemagglutinin (H antigen). Hemagglutinin facilitates
the entry of the virus into cells of the respiratory epi-
thelium, while neuraminidase facilitates the release of
newly produced viral particles (so-called virions) from
infected cells. The inuenza virus has a marked pro-
pensity to mutate its external antigenic composition
to escape the hosts' immune defences. Given this
extreme mutability, the World Health Organisation
(WHO) has introduced a classication of each viral
subtype based on H and N typing. Additionally,
strains are classied on the basis of antigenic type of
the nucleoprotein core (A, B or C), geographical lo-cation of rst isolation, strain serial number and year
of isolation. Every item is separated by a slash mark
(e.g. A/Wuhan/359/95 [H3N2]).
In this century there have been four pandemics
caused by so-called antigenic shift (a major change
in H conguration with or without a concomitant
change in N and perhaps viral alteration of tissue
tropism) leading to the appearance of a new sub-
type against which there is little circulating natural
immunity. Pandemics are thought to originate in
Southern China where ducks (the animal reservoir
and breeding ground for new strains), pigs (whichare thought to be the biological intermediate host
or `mixing vessel') and humans live in very close
proximity [2]. Minor changes in viral antigenic
congurations, known as `drift', cause local or
more circumscribed epidemics. The recently isolated
Hong Kong avian inuenza (A/HK/156/97 [H5N1]
virus appears to be an example of a zoonotic
infection with direct spread of the avian virus to
humans [35]. Pandemics by denition cause a
very high morbidity and mortality burden [6]. The
191819 pandemic is estimated to have caused up
to 40 million deaths world-wide.
1.1. Prevention and early treatment of inuenza
Eorts to prevent or treat inuenza have had their
mainstay in two separate approaches: vaccines and
antivirals (ion channel inhibitors and neuraminidase
inhibitors).
1.1.1. Inuenza vaccines
Current inuenza vaccines are of four types:
1. whole virion vaccines which consist of complete
viruses which have been `killed'; or inactivated, so
that they are not infectious but retain their strain-
specic antigenic properties.
2. subunit virion vaccines which are made of surface
antigens (H and N) only.
3. split virion vaccines in which the viral structure is
broken up by a disrupting agent.4. live vaccines (as yet unlicensed).
The rst three types of vaccines contain the two sur-
face antigens; whole virion and split vaccines also con-
tain antigens which are thought to contribute to a
higher rate of vaccine reactions compared to subunit
vaccines.
Appendix A shows a list of inuenza vaccine produ-
cers and products world-wide, compiled by WHO in
1996 [7].
Periodic antigenic drifts and shifts pose problems
for vaccine production and procurement, as a new
vaccine closely matching circulating antigenic con-guration must be produced and procured for the
beginning of each new inuenza `season'. To achieve
this, WHO has established a world-wide surveillance
system allowing identication and isolation of viral
strains circulating in the dierent parts of the globe.
Sentinel practices recover viral particles from the
naso-pharynx of patients with inuenza-like symp-
toms and the samples are swiftly sent to the labora-
tories of the national inuenza centres (110
laboratories in 79 countries). When new strains are
detected the samples are sent to one of the four
WHO reference centres (London, Atlanta, Tokyo andMelbourne) for antigenic analysis. Information on cir-
culating strains is then sent to WHO, who in
February of each year recommends, through a com-
mittee, the strains to be included in the vaccine for
the forthcoming `season'. Individual governments may
or may not follow WHO recommendations.
Australia, New Zealand and more recently South
Africa follow their own recommendations for vaccine
content.
Surveillance and early identication thus play a cen-
V. Demicheli et al. / Vaccine 18 (2000) 9571030 959
7/31/2019 Demicheli Et Al_2000_Prevention and Early Treatment of Influenza in Healthy Adults
4/74
tral part in the composition of the vaccine.
Traditionally, inuenza vaccines have been targeted to
the elderly and those at serious risk of complications.
Despite clear theoretical advantages in the use of vac-
cines, their uptake has been patchy. Studies in family
practices suggest that 20% is a reasonable estimate of
inuenza vaccine utilisation in the Canadian popu-
lation [810]. The current low level of inuenza vac-
cine uptake in targeted populations may reect
uncertainty on the part of primary care and public
health practitioners and health policy decision-makers
regarding vaccine eectiveness.
One possible reason may be the diversity of regu-
lations for the nancing and reimbursement of the
vaccines. Other reasons may include perceived low
ecacy due to the mutable viral conguration, the
perceived commonality of the disease, which may
breed contempt and, strangely, a misperception of the
burden imposed by the disease on society. Nowhere
is this more marked than in the case of healthy
adults in employment, a population which would
most benet from protection against inuenza.
Epidemics in settings such as schools, barracks, pris-
ons, oces, hospitals and industrial complexes cause
great losses, but are seldom prevented by vaccination
of sta.
Despite the publication over a period of more than
ve decades of a large number of reports of con-
trolled clinical trials, there remains substantial uncer-
tainty about the clinical eectiveness of inuenza
vaccine. This uncertainty is manifested in widely vary-
ing estimates of vaccine eectiveness in the current
health care literature. For example MMWR states:
`The eectiveness of inuenza vaccine in preventing
or attenuating illness varies, depending primarily on
the age and immunocompetence of the vaccine recipi-
ent and the degree of similarity between the virus
strains included in the vaccine and those that circu-
late during the inuenza season. When a good match
exists between vaccine and circulating viruses, inu-
enza vaccine has been shown to prevent illness in ap-
proximately 7090% of healthy persons aged
7/31/2019 Demicheli Et Al_2000_Prevention and Early Treatment of Influenza in Healthy Adults
5/74
be expected when purely clinical outcome measures
are used. This is particularly true when the incidence
of inuenza is low, when the period of observation
extends beyond the usual four to 12 week annual
period of inuenza activity, or when the denition of
illness is imprecise (e.g. respiratory illness). To the
extent that inuenza vaccine modies illness which it
does not prevent, protection might be greater for out-
comes which reect complications of inuenza (e.g.hospitalisation with respiratory illness) rather than
primary infection (e.g. acute respiratory illness). Con-
ventionally, serological diagnosis is based on a four-
fold or greater increase in antibody titre to one or
more virus antigens. There is evidence that vaccinated
individuals are less likely than non-vaccinated persons
to mount an antibody rise following infection with an
inuenza virus antigenically related to strains con-
tained in the vaccine. This phenomenon is thought to
be based, at least in part, on higher pre-infection
antibody titres which result from vaccination. Serolo-
gical methods will therefore `miss' cases of inuenzaamong vaccinated subjects and could be expected to
produce a spuriously high observed protective eect.
Hobson has suggested (without citing supportive evi-
dence) that virus isolation results may be similarly
biased. He proposes that vaccines which fail to pro-
tect against clinical illness may reduce the amount
and duration of virus shedding [15].
Other study design features which might inuence
observed vaccine eectiveness include method of al-
location, extent of blinding and type of virus chal-
lenge (natural or articial). Variability would be
expected to be greater in studies with small samplesizes.
The deciencies of most current and past reviews of
inuenza vaccine eectiveness can be summarised as
follows:
1. lack of comprehensiveness in the identication of
primary studies
2. lack of methodological assessment of primary stu-
dies
3. failure to satisfactorily account for (or in some
cases, to acknowledge) the marked variability in
vaccine eectiveness among controlled studies
4. failure to provide estimates of vaccine eectiveness
under conditions of imperfect antigenic matching
between vaccines and prevalent viruses (that is,
when vaccines contain either a dierent strain or a
dierent subtype of inuenza virus than the preva-
lent virus)
5. lack of credible estimates of vaccine eectiveness in
specic populations currently targeted for inuenza
vaccination (for example, institutionalised elderly,
community-dwelling elderly and persons with under-
lying medical conditions associated with a high risk
of complications [16,17].
These deciencies help to explain discrepancies in
reported vaccine eectiveness in the existing literature.
Moreover, they can be expected to give rise to uncer-
tainty among clinicians and policy-makers regarding
the expected eectiveness of inuenza vaccine in the
population groups for which annual inuenza vacci-nation is currently recommended. In this scenario a
systematic review of the eects of vaccines against
naturally occurring inuenza is necessary to enable de-
cision-makers to devise strategies to deal with inuenza
based on evidence.
1.1.2. Ion channel inhibitor antivirals (Amantadine and
Rimantadine)
The main antiviral compounds used against inu-
enza are amantadine hydrochloride and rimantadine
hydrochloride (amantadine and rimantadine for
short). Amantadine (an anti-Parkinsonism) was intro-duced in the 1950s and found to have antiviral ac-
tivity in 1965. In the USA, amantadine was licensed
for the treatment and prophylaxis of inuenza A/
H2N2 infections by the FDA in 1966 and for pro-
phylaxis and treatment of all inuenza A infections
in 1976. Rimantadine was licenced in 1993 [18]. In
the USA, while amantadine is licensed for treatment
and prophylaxis of adults and children over the age
of one, rimantadine is licensed only for prophylaxis
in children as well as for treatment and prophylaxis
in adults [18]. In the UK amantadine only is licensed
and is administered orally at a recommended does of100 mg a day in healthy adults for ve days (treat-
ment role) or 100 mg a day as long as the risk of
infection lasts (prophylaxis role).
Both compounds interfere with the replication cycle
of type A (but not type B) viruses [19] and are thought
to be ecacious and, given their virus-specic action,
relatively free of adverse eect. Drug resistant H3N2
subtype inuenza A viruses have been isolated during
treatment with amantadine and rimantadine, especially
in institutions, but their clinical signicance is unclear
[20].
Given both drugs' apparent ecacy in both pro-phylactic and therapeutic roles (if administration is
started in time), their relatively scarce use is surpris-
ing [1]. Explanations for this nding include lack of
awareness of the drugs and their properties by medi-
cal practitioners, lack of a rapid diagnostic capability
and concern over their adverse eects, which include
epilepsy. Even more surprising is the list of indi-
cations for use of both drugs. While subjects at high
risk (i.e. subjects with underlying debilitating pathol-
ogies and the elderly) are included, healthy adults, es-
V. Demicheli et al. / Vaccine 18 (2000) 9571030 961
7/31/2019 Demicheli Et Al_2000_Prevention and Early Treatment of Influenza in Healthy Adults
6/74
pecially those working in institutions (such as health-
care workers, nursing home attendants and the mili-
tary) are not. These groups are likely to greatly
benet from the use of the drugs, which could have a
direct impact on length of sickness absence and
diminish considerably the burden of inuenza epi-
demics to society. Additionally a non-systematic
review of the evidence of the ecacy of rimantadine
identied ve small double-blind placebo-controlled
trials of both drugs in a prophylaxis role and nine
trials in a treatment role [21]. The largest study con-
tained 378 individuals, indicating the need to attempt
pooling data to derive more precise estimates of eect
and safety for the compounds. This systematic review
of the eects of amantadine and rimantadine in
healthy adults excludes children, the elderly and indi-
viduals with pre-existing pathologies. However, given
the impact of inuenza in these populations, systema-
tic reviews of the eects of amantadine and rimanta-
dine in children, elderly and at-risk groups should
also be carried out in the future.
1.1.3. Neuraminidase inhibitor antivirals (NIs)
In recent years a new generation of antiviral com-
pounds has been developed and is currently in pre-
registration phase III trials. These compounds, known
collectively as neuraminidase inhibitors (NIs), are:
. Nebulised Zanamivir developed by GlaxoWellcome
PLC (UK).
. Oral Oseltamivir (formerly known as Ro 64-0796
or GS 4104) co-developed by Gilead Sciences Inc.
(Foster City, CA, USA) and Homann-La RocheLtd (Basle, CH). Gilead Sciences Inc. still retains
the intellectual property rights to Oseltamivir.
Zanamivir is a so-called second-generation NI, whereas
Oseltamivir represents the third generation of such
compounds [22]. NIs act by inhibiting the entry of
viral particles into the target cell and subsequent
release of virions from the infected cell, neuraminidase
being essential for both functions. Both Oseltamivir
and Zanamivir appear to be eective against Inuenza
A and B, while amantadine is eective only against
inuenza A.
NIs could be used in both a preventive role and todiminish the severity of the illness [23], to:
. treat infected individuals
. supplement protection against infection in individ-
uals not fully protected by vaccination
. provide protection for individuals unable to receive
vaccine (e.g. individuals allergic to eggs)
. provide short term prophylaxis in family settings
. supplement vaccination during pandemics when vac-
cine stocks may be limited
. control outbreaks in institutions such as nursing
homes or prisons
. control outbreaks in settings such as factories,
oces or the military
. generally interrupt viral transmission.
Homann-La Roche and GlaxoWellcome are target-
ing the registration and marketing of their com-
pounds to the year 2000 inuenza season [24]. AsNIs are likely, if proved eective and safe, to become
a major form of prophylaxis and treatment of inu-
enza, reviewing and updating the available evidence is
necessary to provide an accurate assessment of their
eects.
1.2. Rationale for the economic evaluation
J95, the British Army's ICD-based surveillance sys-
tem, indicated that in soldiers respiratory disease is the
second highest cause of morbidity and sixth highestcause of productivity losses (measured in working days
lost, or WDL) both on world-wide military operations
and when in barracks [25]. Further work carried out
by the Department of Public Health of the University
of Glasgow [26] shows that within the `respiratory dis-
ease' code block approximately 40% of the morbidity
in the 19961997 season was caused by clinical inu-
enza. In some Army subpopulations (such as recruits
undergoing training) the burden of respiratory disease
is much higher (37 attendances per 1000 personnel per
month in Training Establishment compared to 13 at-
tendances per 1000 personnel per month in the rest ofthe Army). Inuenza, then, is an important recurring
public health problem for the British Army, as it
threatens the health and hence eciency of its work-
force, the most important resource that any organis-
ation has at its disposal.
Before embarking in a major expenditure pro-
gramme to purchase large quantities of these inter-
ventions (given that clinical inuenza is a disease of
such high incidence among the military) the Ministry
of Defence of the United Kingdom wanted to make
sure that resources used in the prevention programme
would be recouped by its benets. This provided therationale for an economic evaluation comparing the
costs and eects of each course of action. However,
preliminary work prior to undertaking the evaluation
indicated that there were considerable uncertainties as
to the eectiveness and safety of vaccines, antivirals
and NIs. This provided the main reason for the com-
missioning of three Cochrane reviews [2729] prior to
carrying out the economic evaluation. The evaluation
has been conducted and reported according to the
BMJ guidelines for economic submissions [30].
V. Demicheli et al. / Vaccine 18 (2000) 9571030962
7/31/2019 Demicheli Et Al_2000_Prevention and Early Treatment of Influenza in Healthy Adults
7/74
2. Methods
2.1. Methods for the reviews
2.1.1. Objectives
In comparisons between groups intended for the
interventions and control/placebo groups the following
hypotheses were tested:
2.1.1.1. Cases. There is no dierence in the number of
cases of inuenza and their severity.
2.1.1.2. Adverse eects. There is no dierence in the
number and severity of adverse eects (both systemic
and localised).
2.1.2. Selection criteria
See Table 1.
2.1.3. Search strategy
. A MEDLINE search was carried out using the
extended search strategy of the Cochrane Acute
Respiratory Infections (ARI) Group [31] with the
following search terms or combined sets from 1966
to the end of 1997 in any language: inuenza; route
(oral) OR route (parenteral); vaccine; amantadine;
rimantadine; neuraminidase inhibitors; Oseltamivir ;
GS 4104; Ro 64-0796; Zanamivir
. The bibliography of retrieved articles was examined
in order to identify further trials
. A search was carried out of the Cochrane
Controlled Trials Register (CCTR) and ofEMBASE (199097 for Inuenza Vaccines and for
NIs; 1985 to 1997 for Amantadine and
Rimantadine)
. The journal Vaccine was handsearched from its rst
issue to the end of 1997 [32,33]
. The manufacturers, rst or corresponding authors
of evaluated studies and researchers active in the
eld were contacted in order to locate unpublished
trials.
2.1.3.1. Trial quality assessment. Two reviewers read
all trials retrieved in the search and applied inclusion
criteria. Trials fullling these criteria were assessed for
quality and results analysed by the same authors. Dis-
agreements on trial quality were arbitrated by a third
author. Assessment of trial quality were made accord-
ing to the following criteria:
1. generation of allocation schedule (dened as the
methods of generation of the sequence which
ensures random allocation).
2. measure(s) taken to conceal treatment allocation
(dened as methods to prevent selection bias, i.e. to
ensure that all participants have the same chance of
being assigned to one of the arms of the trial. This
protects the allocation sequence before and during
allocation)
3. number of drop-outs of allocated healthcare worker
participants from the analysis of the trial (dened as
the exclusion of any participants for whatever
reason deviation from protocol, loss to follow-
up, withdrawal, discovery of ineligibility; while the
unbiased approach analyses all randomised partici-
pants in the originally assigned groups regardless of
compliance with protocol, known as intention to
treat analysis)
4. measures taken to implement double blinding (a
double-blind study is one in which observer(s) and/
or subjects are kept ignorant of the group to which
the subjects are assigned, as in an experiment, or of
the population from which the subjects come, as in
a non-experimental situation. Unlike allocation con-cealment, double blinding seeks to prevent ascer-
tainment bias and protects the sequence after
allocation)
For criteria 2, 3 and 4 there is empirical evidence that
low quality in their implementation is associated with
exaggerated trial results [34] and it is reasonable to
infer a quality link between all four items. The four
criteria were assessed by answering a questionnaire; see
Appendix B.
2.1.3.2. Data collection. The following data were
extracted, checked and recorded:
. Characteristics of trials: date; location; setting; case
denitions used (clinical, serological, virological);
surveillance system; type and length of epidemic
(denition used, characteristics of circulating virus);
sponsor (specied, known or unknown); publication
status
. Characteristics of participants: number of partici-
pants; age; gender; ethnic group; risk category; occu-
pation
. Characteristics of interventions: type of intervention;
type of placebo; dose; treatment or prophylaxisschedule; length of follow-up (in days); route of ad-
ministration
. Characteristics of outcome measures:* Numbers and seriousness of inuenza cases (how-
ever dened) occurring in vaccine and placebo
groups. Other outcome measures used to assess
eects included cases of inuenza clinically
dened; cases of inuenza clinically dened on
the basis of a specic list of symptoms and/or
signs; cases of inuenza conrmed by laboratory
V. Demicheli et al. / Vaccine 18 (2000) 9571030 963
7/31/2019 Demicheli Et Al_2000_Prevention and Early Treatment of Influenza in Healthy Adults
8/74
Table1
Selectioncriteriaappliedtoretrievedstudiestodetermineinclusioninsystemat
icreview
Review
Study
criteria
Participants
Interventions
Clinicaloutcomes
Adverseeects
Selectioncriteriacommon
toall
Rando
mised/quasi-
randomiseda
studiesinhumans
compa
redtoplacebo,
controls,
nointervention;orcomparing
types,
doses/schedulesof
intervention
Apparentlyhealthy,r75%
aged14to60
Interventionirrespectiveo
f
viralantigenicconguration
Numbersand/orseverityof
inuenzacases(however
dened)occurringin
interventionandplacebo
groups
Numberandseriousnessof
ad
verseeects
Inuenzavaccines
Protec
tiveeectofinuenza
vaccin
efromexposureto
natura
llyoccurringinuenza
Inuenzaimmunestatus
irrelevant
Attenuated,
killedorlive
vaccinesorfractionsthereof
administeredbyanyroute
Nootherspeciccriteria
Sy
stemiceectsincludeof
malaise,nausea,
fever,
arthralgias,rash,
headacheand
moregeneralisedandserious
sig
ns.Localeectsinclude
induration,
sorenessand
rednessatinoculationsite
(in
jectedvaccines)andrhinitis
an
dsorethroat(inhaled
va
ccines)
Amantadineand
rimantadine
forinuenza
Protec
tionortreatmentof
amant
adineand/or
rimantadinefromexposureto
natura
llyoccurringinuenza
Nootherspeci
ccriteria
Amantadineand/or
rimantadineasprophylaxis
and/ortreatmentforinu
enza
Nootherspeciccriteria
GI(diarrhoea,
vomiting,
dy
spepsia,
nausea,
co
nstipation);increasedCNS
ac
tivity(light-headedness,
co
ncentrationproblems,
insomnia,
restlessness,
ne
rvousness);decreasedCNS
ac
tivity(malaise,depression,
fatigue,vertigo,
feelingdrunk);
sk
in(urticariaandrash)
NIsforinuenza
Protec
tive/treatmenteectof
oralO
seltamivirand/or
Zanam
ivirinnaturallyor
articiallyoccurringinuenza
Nootherspeci
ccriteria
Oseltamivirand/orZanam
ivir
asprophylaxisand/or
treatmentforinuenza
Alsotemporaldistributionof
cases,andotheroutcomese.g.
distributionofsymptoms
underthecurve;timeto
improvement
Localandsystemicadverse
e
ects
a
Astudyisrandomisedwhenit
appearsthattheindividuals(orotherexp
erimentalunits)followedinthestudyweredenitelyorpossiblyassignedprospectiv
elytooneoftwo(ormore)
alternativeformsofhealthcareusin
grandomallocation.
Astudyisquasi-randomisedwhenitappearsthattheindividua
ls(orotherexperimentalunits)followedinthestudyweredenitelyor
possiblyassignedprospectivelytoo
neoftwo(ormore)alternativeformsofh
ealthcareusingsomequasi-randommethodofallocation(suchasalternation,
date
ofbirthorcaserecordnum-
ber).
V. Demicheli et al. / Vaccine 18 (2000) 9571030964
7/31/2019 Demicheli Et Al_2000_Prevention and Early Treatment of Influenza in Healthy Adults
9/74
tests; hospital admissions; complications; working
day lost in episodes of sickness absence regardless
of cause* Adverse eects: presence and type, with local
symptoms presented in the analysis separately
from systemic symptoms; number of withdrawals
due to adverse eects. Individual adverse eects
have been considered in the analysis, as well asa combined endpoint (any or highest symptom).
2.1.3.3. Denitions
Epidemic period. Four dierent denitions of `epi-
demic period' were found:
. the interval between the rst and the last virus iso-
lation in the community
. the interval during which inuenza virus was recov-
ered from more than a stated percentage of ill subjects
. the period during which an increase of respiratory
illness more than a stated percentage was recorded
. the winter period taken as a proxy for epidemic
period.
The data were included regardless of the denition of
epidemic period used in the primary study. When data
were presented for the epidemic period and the entire
follow-up period, those occurring during the former
were considered.
Clinically dened case. A clinically dened case was
assumed as any case denition based on symptoms
without further specication. The specic denitionwas assumed as:
. `u-like illness' according to a predened list of
symptoms (including the CDC case denition for
surveillance)
. `upper respiratory illness' according to a predened
list of symptoms.
When more than one denition was given for the same
trial, data related to the more specic denition were
included.
Laboratory conrmation of cases. The laboratory
conrmation of cases found were:
. virus isolation from culture
. four-fold antibody increase (HI) in acute or conva-
lescent phase sera
. four-fold antibody increase (HI) in post-vaccination
or post-epidemic phase sera.
When more than one denition was given for the same
trial, data related to the more sensitive denition (sero-
conversion) were included.
Hospital admission rates. Hospital admission rates
were calculated as the proportion of cases hospitalised
for respiratory causes.
Complications. Complications were considered as the
proportion of cases complicated by bronchitis, pneu-
monia or otitis.
2.1.3.4. Data synthesis. The relative risks of events
(cases of inuenza, deaths, and adverse eects) com-paring treatment and placebo/control groups from
the individual trials were combined using Mantel-
Haenszel meta-analytical techniques. We did not com-
bine estimates from treatment and prophylactic trials
as these were conducted to answer dierent study
questions. Between-trial variability in results was
examined and incorporated into the estimates of
uncertainty of treatment eect using random eects
models where appropriate. In treatment trials the
choice of methods for combining the estimates of
severity of inuenza depended on the format in
which the data was presented. Where possible, com-parisons were made between the mean duration of
symptoms in the two groups, and methods for com-
bining dierences in means were used. Specically,
where the data were presented as the number of sub-
jects with duration of symptoms beyond a cut-o
time period these were presented as `Cases with fever
at 48 h'. The bewildering array of outcomes used in
the treatment trials (see Results section) prevented us
from using more than the `cases with fever' outcome.
Included trials did not contain sucient information
to enable us to assess the number of cases with no
documented fever at entry into the trial.For the vaccine trials, separate analyses were per-
formed for live aerosol vaccines, inactivated parent-
eral vaccines and inactivated aerosol vaccines.
Clinical inuenza outcomes were specied according
to whether specic criteria were or were not used, for
which estimates were produced separately, and com-
bined (where trials reported both denitions, only the
wider denition was retained for analysis). Vaccine
ecacy was estimated by calculating the common
relative risk, using the Mantel-Haenszel method (xed
eect model) when the trial results were consistent, or
the DerSimonian and Laird method (random eectsmodel) when signicant heterogeneity was evident
between the study results. Between-study heterogen-
eity is to be expected in vaccine trials as there are
unpredictable systematic dierences between trials in
circulating strains and levels of local immunity. Once
the relative risk (RR) had been obtained, vaccine e-
cacy (VE) was calculated as VE=1-RR. Similar ana-
lyses were also undertaken for other events, such as
complications, hospital admissions and adverse
eects.
V. Demicheli et al. / Vaccine 18 (2000) 9571030 965
7/31/2019 Demicheli Et Al_2000_Prevention and Early Treatment of Influenza in Healthy Adults
10/74
In addition to the traditional estimate of vaccine e-
cacy, the eect of vaccination on the number of clini-
cal cases was estimated by averaging the risk
dierences (inuenza rate in vaccinated group minus
inuenza rate in control group). Where the total num-
ber of clinical inuenza cases depends more on the
number of other inuenza-like illnesses than true inu-
enza A illnesses, it is more likely that an intervention
will appear to reduce the total number of cases by an
absolute amount (i.e. a constant risk dierence) than
by a relative amount (i.e. a constant relative eect).
As the data on average time o work was reported
as a continuous measurement, these results were
expressed as dierences in means, and combined using
the weighted mean dierence method. Caution should
be exercised in interpreting these results as the data are
very skewed.
Several trials included more than one active vaccine
arm. Where several active arms from the same trial
were included in the same analysis, the placebo group
was split equally between the dierent arms, so thatthe total number of subjects in any one analysis did
not exceed the actual number in the trials.
2.1.4. Methods for the economic evaluation
2.1.4.1. Evidence-based alternative interventions to mini-
mise the burden of inuenza. While the three Cochrane
reviews were underway, we assumed a hypothetical
scenario in which all available means had a preventive
and treatment impact on inuenza. We also considered
it likely that such means would produce adverse eects
and have clinical outcomes not homogeneous for qual-ity of life. In this case, the alternatives to be explored
would be:
. which is the best single alternative
. which is the best combination of alternatives
. which is the best combination of alternatives
depending on the outcome measure considered
(avoided cases, quality weighted avoided cases,
severity of avoided cases, hospital admissions
avoided and working days lost (WDL)).
We aimed to compare these alternatives with the cur-
rent Army policy on inuenza prevention (do-noth-ing).
Once the reviews had been completed, the results led
us to introduce considerable changes to our compara-
tors. The changes (with the reasons in brackets) are
summarised in Table 2.
For our evaluation we chose the viewpoint of the
funder, the MOD (UK). We thus focused on the
eects of preventing inuenza in MOD/Army person-
nel although we believe that our methods are equally
applicable to populations of employed healthy adults,
especially in an epidemic situation. These would
include emergency services and employees of compa-
nies producing essential goods and services.
We were able to test the eect of this assumption by
setting our results in the context of a distribution of
similar variables derived from our widely known and
recently updated systematic review of the economics of
inuenza [6,35,36].
We attempted to incorporate into our evaluation
individual soldier preferences for the possible preven-
tive means. One of the eects of adopting the view-
point and decision-making perspective of the MOD/
Army was the possibility of incorporating the inu-
enza preventive campaign into existing immunisation
and routine procedures at no incremental administra-
tive cost. However, in the sensitivity analysis we have
used administration costs derived from the ratio `vac-
cine cost/total administration cost' calculated from
our systematic review of the economics of inuenza
[6,35,36].
Final selection of alternatives. Our nal criteria forthe choice of alternatives were:
. evidence of ecacy;
. evidence of safety;
. practicality of organisational implementation in the
setting of the British Army.
On the basis of the rst criterion all remaining
alternatives in the third column of Table 2 are prac-
ticable and acceptable. However applying the other
two criteria and assuming an average inuenza epi-
demic period of 46 days (as in the trials included in
the reviews) the alternatives of oral amantadine, oralrimantadine and oral Oseltamivir are no longer prac-
ticable. It is very unlikely that whole bodies of sol-
diers would comply with the requirement of
protracted daily oral drug schedules. This assumption
was further conrmed by the nding of our prefer-
ence time trade-o exercise (Table 4) in which sol-
diers preferred the risk of contracting inuenza to
that of experiencing adverse eects such as nausea or
gastrointestinal disturbances.
Two other factors contribute to making the preven-
tion of inuenza with antimicrobials and NIs proble-
matic. Firstly it is doubtful whether the protractedlogistical eort involved in maintaining the chemopro-
phylaxis campaign for 46 days is feasible. Secondly the
level and timeliness of the information required to
determine with any certainty the `beginning' and the
`end' of the inuenza epidemic is unlikely to be avail-
able, especially when the Army is deployed in dierent
areas of the UK and abroad.
Description of alternatives. Whereas before the
Cochrane review results our provisional decision tree
was very complicated, comprising preventive and treat-
V. Demicheli et al. / Vaccine 18 (2000) 9571030966
7/31/2019 Demicheli Et Al_2000_Prevention and Early Treatment of Influenza in Healthy Adults
11/74
ment alternatives, the nal tree consists only of the
three preventive arms each with inuenza cases with or
without adverse eects.
Form of the economic model. On the basis of the
above considerations we dened an economic model
based on the cost per avoided case to dene the best
preventive strategy, and the cost per avoided case
weighted by individual preference to dene the choiceof the best combination of interventions. The cost
per avoided case was calculated by dividing the total
costs of the interventions by the number of cases
avoided.
Data collection and assumptions made. We based our
model on a set of assumptions, which are summarised
in Table 3.
The variables and the ranges across which we car-
ried out our sensitivity analysis together with the
rationale are summarised in Table 4.
3. Results
3.1. Results of the reviews
3.1.1. Description of studies
Identied trials are listed and described in the table
of included studies using the name of the rst author
and the publication year; see Appendix C. A list and
description of excluded studies (with reason for exclu-
sion) is available from the authors.
3.1.1.1. Inuenza vaccines. The tables of comparisons
were constructed according to the following criteria
(Fig. 1):
1. Inuenza vaccine versus placebo* All studies comparing any inuenza vaccine
against a placebo (inert substances or non
Table 2
Possible alternatives to prevent and treat inuenza, before and after reviews of the evidence
Items Before Cochrane reviews After Cochrane reviews
Which is the best single alternative for
prevention
Oral vaccines Parenteral vaccines
Aerosol vaccines Oral Amantadine
Parenteral vaccines Oral Rimantadine
Oral Amantadine Oral Oseltamivir
Oral Rimantadine (aerosol/oral vaccines are less eective, or dierences are
minimal and do not currently represent a real alternative.
Zanamivir trials only apparently included laboratory
conrmed outcomes)
Inhaled Zanamivir
Oral Oseltamivir
Which is the best single alternative for
treatment
Oral Amantadine None (all compounds shortened duration of illness by 0.5
days)
Oral Rimantadine
Inhaled Zanamivir
Oral Oseltamivir
Which is the best combination of
alternatives
Prevention only Prevention only
Treatment only
Prevention treatment
Outcome measure Laboratory cases Laboratory cases
Clinical cases Clinical cases
WDL (the prevention of clinical cases is the only public health
target. Not enough outcome data were presented in the trials
to include any other outcomes)
Hospital admissions
Deaths
Complications
Length of epidemics (i.e. required
duration of antiviral & NI preventive
treatment)
84 days (SD=33.6) according to
Communicable Disease Reports
`Inuenza Surveillance England
and Wales' (199197)
62 days (SD=27) (according to inuenza vaccines trials
included in the Cochrane review [27])
V. Demicheli et al. / Vaccine 18 (2000) 9571030 967
7/31/2019 Demicheli Et Al_2000_Prevention and Early Treatment of Influenza in Healthy Adults
12/74
inuenza vaccines) were included in this
group* Subgroup analysis were carried out for live aero-
sol vaccine, inactivated parenteral vaccine, and
inactivated aerosol vaccine* The parenteral route comprised both intramuscu-
lar and subcutaneous route* Dierent dosages and schedules of the vaccine
and the presence of dierent adjuvants were not
compared; and data from arms of trials compar-
ing only vaccine composition or dosage were
pooled in the analysis.
2. At least one vaccine strain recommended for that
year (as an indicator of goodness of serological t)
versus placebo or other vaccines:* All trials in which the studied vaccine contained
at least one of the A strains recommended for
that year by WHO or single governments (WHO
recommendations were published since 1973 only)
were included, independently from substances
used in the control arm* Subgroup analysis was carried out according to
control group for the recommended vaccine
against placebo, against inuenza B vaccine, and
against other non recommended A strains* Vaccines containing only a B recommended strain
were excluded from this comparison since a num-
ber of authors used monovalent B vaccine as pla-
cebo in the control arm which may generate
confusion* The compliance of the study vaccine with the o-
cial recommendations was checked by reviewing
WHO records when possible. In case of ambigu-
ity (in the oldest trials), the opinion stated by
authors was taken into account* The compliance of a live attenuated vaccine with
the recommendation has been decided according
to the antigenic comparability to the wild strains
3. Vaccine matching circulating strain versus placebo
or other vaccines:* All trials in which the studied vaccine contained
the strain matching the circulating virus (or at
least one of several circulating viruses) were
included in this group of comparison, indepen-
dently from substances used in the control arm* Subgroup analysis was carried out according to
the control group; matching vaccine against pla-cebo, against inuenza B vaccine, and against
other non recommended A strains* In cases of an incomplete match or ambiguity of
wording, the opinion stated by authors was also
taken into account. Minor viral drift clearly sta-
ted was assumed as non-matching.
Twenty papers describing 39 trials of sub-trials were
identied. Some of them had more than two arms,
comparing dierent vaccines, routes of administration,
Table 3
Basic assumptions of the model
Variable Assumption Source/Rationale
Population Army eectives as at 1 August 1998 Defense Analytical Services Agency (DASA)
Gender and age dierences Only incidence dierences will be tested in
sensitivity analysis
DASA
Incidence of inuenza Sickness rates for inuenza in 1997 DASA
Eectiveness Meta-analysis estimate of RCTs using inuenza
clinical outcomes
Cochrane Reviews [2729]
Adverse eects Frequent symptom reported in RCTs included
in Cochrane reviews comparable across range of
preventive interventions
Cochrane Reviews [2729]
Individual preferences Mean score of preferences expressed as
combination of category rating and time-trade-
o
Study on a sample of 40 soldiers
Preventive intervention costs Acquisition costs Defense Medical Supply Agency and authors'
assumption (NI)
Duration of treatment for antivirals and NIs Mean duration of inuenza epidemics from
vaccines RCTs
Cochrane Review [27]
Preventive intervention administration costs Nil Interviews with medical commanders
Productivity losses due to inuenza Nil Preventive interventions do not have dierent
eects under this perspective
Productivity loss from adverse intervention
events
Nil Diculties in valuation
V. Demicheli et al. / Vaccine 18 (2000) 9571030968
7/31/2019 Demicheli Et Al_2000_Prevention and Early Treatment of Influenza in Healthy Adults
13/74
schedules or dosages. These trials were split into sub-
studies. Some trials took in account the history of pre-
vious vaccine immunisations.
Included trials assessed three kinds of vaccine: live
attenuated aerosol, inactivated aerosol and inactivated
parenteral. Four trials of live attenuated vaccine were
included, all placebo controlled. These involved 26,369
subjects. The mean treatment size was 2028 individuals
(median 999, 25th percentile 508, 75th percentile 1071),
and the mean placebo arm size was 1739 (median 508,
25th percentile 289, 75th percentile 547 individuals).
Two studies which assessed inactivated vaccine aerosol
were included. Both were placebo controlled and
involved 1506 subjects. The mean treatment size was
335 individuals (median 333, 25th percentile 195, 75th
percentile 473), and the mean placebo arm size was 42
(median 42, 25th percentile 24, 75th percentile 59 indi-
viduals).
Most studies assessed ecacy of inactivated parent-
eral vaccines against placebo or other inuenza vac-
Table 4
Assumptions tested in the sensitivity analysis
Variable Assumption Source/Rationale
Incidence of i n uenza Range of values from basic Army rate to training
regiment rate
DASA, J97 & Glasgow University study [25,26]
Eectiveness Range of estimate from meta-analysis of RCTs using
inuenza clinical outcomes case denition and from
meta-analysis of RCTs using laboratory-based
inuenza case denition
Cochrane Reviews [2729]
Eectiveness Range of 95% Condence intervals around eect on
outcome (clinical case denition)
Cochrane Reviews [2729]
Adverse eects Range of incidence estimates with arbitrary variation Authors' assumption
Individual preferences 25th and 75th percentiles scores of preferences
expressed as a combination of category rating and
time-trade-o
Study on a sample of 40 soldiers
Duration of treatment for antivirals and
NIs
Minimum and maximum duration of inuenza
epidemics from vaccines RCTs
Distribution of duration is symmetrical.
Cochrane Review [27]
Vaccines administration costs Ratio of administration to vaccine costs derived from
economic studies on vaccines
Systematic review of economic studies [6,36]
Antivirals and NIs administration costs Arbitrary ratio of administration to drug costs Authors' assumptions
Fig. 1. Summary of inuenza vaccines in healthy adults (95%CI=95% Condence intervals).
V. Demicheli et al. / Vaccine 18 (2000) 9571030 969
7/31/2019 Demicheli Et Al_2000_Prevention and Early Treatment of Influenza in Healthy Adults
14/74
cines (some of them used a monovalent inuenza B
vaccine as placebo). They involved 23,628 subjects.
The mean treatment size was 550 individuals (median
432, 25th percentile 161, 75th percentile 920), and the
mean placebo arm size was 358 (median 311, 25th per-
centile 66, 75th percentile 518 individuals).
Surveillance methods were prospective or retrospec-
tive, active (by phone interview or questionnaire com-
pilation) or passive (ill subjects spontaneously
presenting). Mean length of follow up was 87 days
(median 79 days, 25th percentile 61 days, 75th percen-
tile 119 days).
The duration of the epidemic was specied by 17
trials. Mean length of the epidemic period was 62 days
(median 63 days, 25th percentile 42 days, 75th percen-
tile 77 days).
3.1.1.2. Amantadine and Rimantadine
Preventive trials. Seventeen preventive trials met the
inclusion criteria. No unpublished trials were ident-
ied, despite receiving nine letters and three electroniccommunications from manufacturers, authors and
researchers.
The mean amantadine arm size was 494 individuals
(median 151, 25th percentile 97, 75th percentile 348),
the mean rimantadine arm size was 107 (median 108,
25th percentile 92, 75th percentile 122 individuals) and
the mean placebo arm size was 373 individuals (me-
dian 140, 25th percentile 99, 75th percentile 269). The
mean total population was 596 individuals (median
308, 25th percentile 225, 75th percentile 536). The
mean length of follow up was 28 days (median 30
days, 25th percentile 18 days, 75th percentile 42 days).Treatment trials. Ten published treatment trials were
identied. No unpublished trials were identied. The
mean amantadine arm size was 91 individuals (median
72, 25th percentile 15, 75th percentile 110), the mean
rimantadine arm size was 61 (median 56, 25th percen-
tile 15, 75th percentile 104 individuals) and the mean
placebo arm size was 77 individuals (median 76, 25th
percentile 14, 75th percentile 99). The mean total
population was 161 individuals (median 153, 25th per-
centile 30, 75th percentile 225). Mean length of follow
up was 25 days (median 25.5 days, 25th percentile 16
days, 75th percentile 33 days).
3.1.1.3. Neuraminidase inhibitors
Preventive trials. As at 1 January 1999 four accessi-
ble preventive trials met our inclusion criteria. A
further two preventive trials of Zanamivir in abstract
format were identied (Calfee H68 and Monto).
Further data was requested from the manufacturers,
GlaxoWellcome, to allow the inclusion of data from
the trials in the review. GlaxoWellcome provided the
data as requested. The mean Zanamivir arm size was
136 individuals (median 34, 25th percentile 25, 75th
percentile 61), the mean Oseltamivir arm size was 1040
(median, 25th percentile and 75th percentile 1040 indi-
viduals) and the mean placebo arm size was 189 indi-
viduals (median 21, 25th percentile 9, 75th percentile
397). The mean total population was 475 individuals
(median 68, 25th percentile 36, 75th percentile 853).
Mean length of follow up was 11 days.
Treatment trials. As at 1 January 1999 three accessi-ble treatment trials were identied which fullled the
inclusion criteria. A further two treatment trials of
Zanamivir in abstract format were also identied.
Despite a request to the manufacturers, GlaxoWell-
come did not release more detailed data in time for in-
clusion in the review.
The mean Zanamivir arm size was 80 individuals
(median 43, 25th percentile 25, 75th percentile 43), the
mean Oseltamivir arm size was 920 (median 1040, 25th
percentile 920, 75th percentile 1040 individuals) and
the mean placebo arm size was 107 individuals (me-
dian 85, 25th percentile 22, 75th percentile 151). Themean total population was 315 individuals (median
243, 25th percentile 65, 75th percentile 449). Mean
length of follow up was ve days.
Preventive and treatment trials. Only one trial was
identied containing both preventive and treatment
interventions.
3.1.2. Methodological quality of included studies
Two reviewers assessed allocation method, allo-
cation concealment, blinding and completeness of fol-
low-up.
3.1.2.1. Inuenza vaccines. There were 20 trials in all,
13 of which were placebo controlled. Three trials used
an inuenza B vaccine in the control arm, considering
it as a placebo. Four trials compared two or more
inuenza vaccines but did not use a control arm.
Thirteen trials reported data on adverse eects, but
only seven were included in the analysis: one did not
have sucient reporting and ve trials did not have a
placebo arm. The overall quality of the trials was
good.
Assessed allocation concealment was adequate in 12
of the trials, inadequate in six and unclear in two.Fifteen trials were properly randomised, four stated
that the allocation method was quasi-random, and one
trial did not report information about randomisation.
Assessment was double blinded in 14 trials. Two trials
were single blinded and four did not mention blinding.
Two studies were eld trials.
3.1.2.2. Amantadine and Rimantadine. There were 27
trials in all, 26 of which considered either amantadine
and/or rimantadine ecacy and one which considered
V. Demicheli et al. / Vaccine 18 (2000) 9571030970
7/31/2019 Demicheli Et Al_2000_Prevention and Early Treatment of Influenza in Healthy Adults
15/74
adverse eects only. Eleven preventive trials and seven
treatment trials reported sucient data on adverse
eects. The quality of preventive and treatment trials
is discussed separately.
Preventive trials. The quality of the preventive trials
was relatively good, considering the age of the trials.
Among the 17 preventive trials, 15 stated that the al-
location method was randomisation, although only
four mentioned a particular method and two did not
mention random allocation at all. These two trials
have therefore been classied as controlled clinical
trials (CCTs) rather than RCTs. All preventive trials
were stated to be double blind with the exception of
Payler which was open and had no placebo group
(the comparison group was no intervention other
than inuenza vaccine at the beginning of the sea-
son).
Treatment trials. Among the 10 treatment trials, nine
stated that the allocation method was randomisation;
no trials mentioned a particular method; and one(Hornick) did not mention random allocation at all.
Major aws in the reporting of trials lay in the follow-
ing:
. Lack of information on the completeness of follow-
up. In many trials there was a large dierence
between the number randomised and the number
who actually participated
. Lack of detailed description of methods to conceal
allocation, with many trials just describing a `double
blind' procedure
. Frequent inconsistencies in the reporting of numer-
ators and denominators in various arms of trials
. In the treatment trials, the use of a bewildering
variety of outcomes, such as severity scores, of
which none are alike. This makes the task of
meta-analysis impossible and leads to a great loss of
information.
3.1.2.3. Neuraminidase inhibitors. Overall methodologi-
cal quality appeared good, in keeping with the mainly
early report nature of the results of the clinical trials
of such potentially important compounds. However,
detailed descriptions of methods and steps taken to
ensure allocation concealment were not specic, lead-
ing us to grade this aspect of the trials `unclear'. This
is potentially a very important point when dealing
with cases of self-limiting upper respiratory tract infec-
tions with or without systemic symptoms, in which the
potential for a placebo eect is great. Additionally as
some trials (the WV series for instance) relied on clini-
cal case denitions the potential for bias (and overesti-
mation of eect) is even greater.
3.2. Eects of inuenza vaccines
3.2.1. Eect of vaccination on clinical cases of inuenza
Trial data for the two denitions of inuenza (no
case denition and specic case denition) are pre-
sented separately for each of the three types of vaccine:
live aerosol, inactivated parenteral and inactivated
aerosol. Signicant heterogeneity was detected between
trial results for most comparisons, and the gures
quoted are estimated from random eects models.
The live aerosol vaccines were not eective for cases
of either denition. A combined analysis of data from
the two trials estimated the vaccine ecacy to be 2%
(95%CI: 58%).
The inactivated vaccines did oer signicant protec-
tion. Taking the data from the 10 trials together,
regardless of case denition, the parenteral vaccine
reduced the number of cases by 29% (95%CI: 12
42%). The ecacy of the inactivated aerosol vaccine
was higher for the unspecied case denition
(VE=31%, 95%CI: 551%) but not the specic inu-enza case denition (VE=26%, 95%CI: 145%).
The estimates of ecacy were more consistent when
the treatment eect was expressed as a risk dierence
rather than a relative eect. Estimation as risk dier-
ences suggest that 5% (95%CI: 28%) and 9%
(95%CI: 316%) fewer participants experienced inu-
enza like illnesses who received inactivated parenteral
vaccine and inactivated aerosol vaccine respectively.
3.2.2. Eect of vaccination on serologically conrmed
cases of inuenza
Data from two studies showed that aerosol live vac-cines reduced the number of serologically conrmed
cases of inuenza by 79% (95%CI: 4492%). Six stu-
dies provided data for inactivated parenteral vaccines,
showing a similar ecacy of 65% (95%CI: 4479%).
No studies of inactivated aerosol vaccine reported
numbers of serological conrmed cases.
3.2.3. Eect of vaccination on other outcomes
Three trials of parenteral inactivated vaccine evalu-
ated time o work, estimating that vaccination saved
on average around 0.4 working days. This result was
nearly statistically signicant. Hospital admissionswere also lower, but not statistically signicant. There
was little dierence in complication rates between vac-
cinated and unvaccinated groups.
3.2.3.1. Adverse eects aerosol live vaccines. Whilst
signicantly more recipients experienced sore throats
after vaccine administration than placebo adminis-
tration (relative rate=2.5, 95%CI: 1.54.2), the overall
number of local adverse eects was not signicantly
dierent between vaccine and placebo groups. There
V. Demicheli et al. / Vaccine 18 (2000) 9571030 971
7/31/2019 Demicheli Et Al_2000_Prevention and Early Treatment of Influenza in Healthy Adults
16/74
was also no signicant increase in systemic side eects,
although rates of fever and myalgia were higher in vac-
cine than placebo groups. Overall 26% of vaccine reci-
pients reported the combined endpoint for local
reactions, whilst only 8% reported the combined end-
point for systemic eects.
3.2.3.2. Adverse eects inactivated vaccines. Local
tenderness and soreness was more than twice as com-mon among parenteral vaccine recipients than those in
the placebo group (relative rate=2.1, 95%CI: 1.43.4).
There were also increases in erythema (non-signicant),
but not in duration of or arm stiness. The combined
local eects endpoint was signicantly higher for those
receiving the vaccine (relative rate=2.6, 95%CI: 1.6
4.2), with 69% reporting some eect.
None of the systemic eects were individually more
common in parenteral vaccine recipients than placebo
recipients. However the combined endpoint was
increased, and nearly statistically signicant, with 26%
vaccine recipients reporting some side eect than pla-cebo recipients (95%CI: 059%). Overall 30% of
those receiving the vaccine reported possible systemic
eects, although many of these equally could be attrib-
uted to inuenza-like illnesses.
None of the trials on inactivated aerosol vaccines
reported side-eects that could be included in the
analysis. The two studies which evaluated these vac-
cines included parenteral components using an inu-
enza B control group so that the side-eects of the
oral vaccine could not be estimated separately.
3.2.4. Recommended vaccinesSixteen trials evaluated the eect of the vaccines rec-
ommended (by WHO or single governments) on clini-
cal cases of inuenza. Nine of these trials were placebo
controlled, ve made comparisons with inuenza B
vaccine, and two compared recommended inuenza A
vaccines with non-recommended inuenza A and B
vaccines. An additional trial (Tannock) only reported
serologically conrmed cases.
Live aerosol, inactivated parenteral and inactivated
aerosol all had similar vaccine ecacies, although the
estimate for inactivated aerosol vaccines was only
based on the results of one trial. Live aerosols had avaccine ecacy of 13% (95%CI: 520%), inactivated
parenterals an ecacy of 24% (95%CI: 1532%), and
inactivated aerosols an ecacy of 40% (95%CI: 13
59%). Combining the data from all three vaccine types
from the placebo controlled trials, the overall estimate
of vaccine ecacy was 24% (95%CI: 1433%). The
estimate decreased to 22% when the non-placebo con-
trolled trials were included (Fig. 2).
Again, the individual study results were more con-
sistent when expressed as risk dierences than relative
eects. Overall the percentage of participants experien-
cing clinical inuenza decreased by 5% (95%CI: 3
7%) using data from the placebo controlled trials. The
reductions were 3%, 5% and 9% for the live aerosol,
inactivated parenteral and inactivated aerosol vaccines
respectively, the rst gure not being statistically sig-
nicant.
There were signicant reductions in serologically
conrmed cases of inuenza for live aerosol and inacti-
vated parenteral preparations. There was no data for
inactivated aerosol vaccines. Vaccine ecacy was esti-
mated as 48% (95%CI: 2464%) for live aerosol vac-
cines, and 68% (95%CI: 4979%) for inactivated
parenteral vaccines.
3.2.5. Vaccine matching the circulating strain
The highest estimates of vaccine ecacy come from
the analyses of vaccines which were shown to match
the circulating vaccine strain. Twelve trials were
included in these analyses, and seven were placebo
controlled. Since several studies had more than two
arms, the ecacy of the vaccines containing the match-
ing strain was compared against non-matching A or B
inuenza vaccines. None of the live aerosol vaccines
used in the trials matched circulating strains.
Estimates of the ecacy of both parenteral and
aerosol inactivated vaccines in reducing cases of clini-
cal inuenza were similar. Overall the vaccine ecacy
based on results of the placebo controlled trials was
37% (95%CI: 1852%). The estimate declined to 31%
when the non-placebo controlled trials were included.
Expressing the ecacy as a risk dierence, on average
7% (95%CI: 410%) fewer participants who receivedmatched vaccine suered inuenza like illnesses com-
pared to placebo recipients.
The eect of the matched vaccine on serologically
conrmed cases was also larger than in any other
analysis. Overall the results of seven trials reporting
serologically conrmed cases estimated the vaccine e-
cacy to be 72% (95%CI: 5483%).
3.3. Eects of amantadine and rimantadine
All trials tested the eects of amantadine and riman-
tadine on a wide variety of inuenza A viruses. None
tested the eects on inuenza B, on which the mol-
ecules are known to be ineective. Also, no trial tested
the role of the compounds on workplace outbreak con-
trol, which is a pity considering the trial settings (pris-
ons, factories, schools, barracks).
Some trials are likely to have included individuals
who took aspirin to relive symptoms (especially in the
V. Demicheli et al. / Vaccine 18 (2000) 9571030972
7/31/2019 Demicheli Et Al_2000_Prevention and Early Treatment of Influenza in Healthy Adults
17/74
Fig. 2. At least one vaccine recommended for that year compared to placebo or other vaccine in in uenza cases clinically dened
(Expt=experimental arm, ctrl=control arm; 95%CI=95% Condence intervals).
V. Demicheli et al. / Vaccine 18 (2000) 9571030 973
7/31/2019 Demicheli Et Al_2000_Prevention and Early Treatment of Influenza in Healthy Adults
18/74
treatment trials). However the eects of this potential
confounder should have been eliminated by the process
of randomisation.
All trials commenced administration of the com-
pounds within a reasonable time lapse. When the
results of surveillance made it reasonable to do so,
treatment started at the latest 48 h after positive identi-
cation of the rst case in the population and preven-
tion.
Six main comparisons were carried out:
1. Comparison A oral amantadine compared to
placebo in inuenza prevention
2. Comparison B oral rimantadine compared to
placebo in inuenza prevention
3. Comparison C oral amantadine compared to
oral rimantadine in inuenza prevention
4. Comparison D oral amantadine compared to
placebo in inuenza treatment
5. Comparison E oral rimantadine compared to
placebo in inuenza treatment6. Comparison F oral amantadine compared to
oral rimantadine in inuenza treatment
Two minor comparisons, G and H, were also carried
out, each based on the results of a single trial.
For comparisons A, B and C the eects on `cases'
were analysed, stratied either on the basis of clinical-
laboratory criteria (a dened set of signs and symp-
toms backed up by serological conrmation and/or
isolation of inuenza virus from nasal uids) or clini-
cal criteria alone. The eects of amantadine/rimanta-
dine administration on asymptomatic cases (dened
only by serology or viral isolation) were not assessed,as these are of little public health interest.
Comparisons were stratied on the basis of whether
participants had received vaccination or not.
Finally, the adverse eects in the comparisons were
assessed. The `all adverse eects' category includes all
types and was derived from those trials which either
did not report sucient information to allow a more
detailed classication or that presented aggregate data.
Adverse eects incidence is reported in the meta-analy-
sis as event per person, thus the incidence should not
be added as more than one adverse event is likely to
have taken place in the same individual during thetrial. The dierence in incidence of adverse eects is of
importance, rather than the estimated incidence itself,
as the adverse eects reported with these drugs are
very similar to the clinical manifestations of inuenza
infection.
Apart from these caveats the analysis shows that all
types of adverse events were signicantly more likely
to happen when individuals were given amantadine
rather than placebo (with the exception of the `other'
category) but none were signicantly more likely to
take place in the rimantadine or placebo arms. Overall
both drugs appear to be eective and well tolerated,
although the evaluation of the eects of rimantadine
was carried out on a very small study population.
In all comparisons duration of action and protection
appeared directly related to duration of prophylaxis or
treatment with amantadine and rimantadine. This nd-
ing is in keeping both with the half-life of the com-
pounds which are excreted by the kidneys (at the rate
of 6.4 ml/min/kg for amantadine and 1.2 ml/min/kg
for rimantadine [37]) and the self-limiting duration of
the illness. No trials assessed onset of resistance to the
drugs although this is known to be of relatively short
induction time (1027% of patients secrete drug-resist-
ant virus within 45 days of commencing treatment
[37]).
3.3.1. Comparison A oral amantadine compared to
placebo in inuenza prevention
3.3.1.1. Ecacy. Amantadine: 61% (95%CI: 5169%)
ecacious (RR 0.39 95%CI: 0.310.49) in prevent-
ing clinically and laboratory dened inuenza cases;
23% (95%CI: 1134%) ecacious (RR 0.77 95%CI:
0.660.89) in preventing clinically dened inuenza
cases (Fig. 3). There was a signicant variation in the
trial results for the second outcome.
3.3.1.2. Adverse events. All categories of adverse eects
were signicantly more common in participants who
received amantadine than placebo, except for dermato-
logical changes. Nearly twice as many amantadine reci-
pients experienced both increased or decreased CNS
eects, and more than twice as many withdrew from
the trials due to adverse eects (Table 5).
3.3.2. Comparison B oral rimantadine compared to
placebo in inuenza prevention
3.3.2.1. Ecacy. Rimantadine: 72% (95%CI: 8
92%) ecacious (RR 0.28 95%CI: 0.081.08) in
preventing clinically and laboratory dened inuenza
cases; 35% (95%CI: 2065%) ecacious (RR 0.65
95%CI: 0.351.20) in preventing clinically inuenzacases (Fig. 4). The signicance of these ndings
depends on whether a xed or random eect model is
used.
3.3.2.2. Adverse events. Rimantadine recipients were
also more likely to experience adverse eects than pla-
cebo recipients. However, there was no evidence of an
increase in CNS-related eects with rimantadine and
withdrawal rates were similar in both groups (Table
6).
V. Demicheli et al. / Vaccine 18 (2000) 9571030974
7/31/2019 Demicheli Et Al_2000_Prevention and Early Treatment of Influenza in Healthy Adults
19/74
3.3.3. Comparison C oral amantadine compared to
oral rimantadine in inuenza prevention
3.3.3.1. Ecacy. There was no evidence of a dierence
in ecacy between amantadine and rimantadine,
although the condence interval is quite wide (RR
amantadine vs rimantadine 0.88. 95%CI: 0.481.63).In some cases (Plesnik) data on cases of inuenza
have been included both under clinically and serologi-
cally dened, so the two outcomes should not be
added.
3.3.3.2. Adverse events. CNS adverse eects and with-
drawal from trials was more signicantly common
among amantadine recipients than rimantadine recipi-
ents (CNS eects; RR 2.58, 95%CI: 1.544.33; with-
drawals RR 2.30, 95%CI: 1.234.30).
Thus rimantadine appears no less ecacious but
safer than amantadine in preventing cases of inuenza
in healthy adults, although the study sizes of the safety
trials of rimantadine are considerably smaller than
those of amantadine.
3.3.4. Comparison D oral amantadine compared to
placebo in inuenza treatment
3.3.4.1. Ecacy. Amantadine signicantly shortened
duration of fever compared to placebo (by 1.00 days
95%CI: 0.731.29). The meta-analysis is based on
506 subjects (230 in the amantadine and 276 in the
placebo arm). Where time to fever clearance data were
not available (van Voris and Wingeld), a dichoto-
mous outcome was used (cases with fever at 48 h).
Fig. 3. Oral amantadine compared to placebo in inuenza prevention: inuenza cases clinically dened (Expt=experimental arm, ctrl=control
arm; 95%CI=95% Condence intervals).
Table 5
Comparison A: Incidence of adverse eects expressed as a percentage of participants
Amantadine (%) Placebo (%) No. of trials N Signicant
All adverse eects 14.7 10.4 6 4274 Yes
GI eects 5.1 2.4 5 3336 Yes
Increased CNS activity (excitation) 7.5 4.7 9 5002 Yes
Decreased CNS activity (depression) 8.6 7.1 6 3782 Yes
Skin 1.1 6.8 4 918 No
V. Demicheli et al. / Vaccine 18 (2000) 9571030 975
7/31/2019 Demicheli Et Al_2000_Prevention and Early Treatment of Influenza in Healthy Adults
20/74
Amantadine was shown to be signicantly better than
placebo (Peto relative risk 0.21 95%CI: 0.070.66).
3.3.4.2. Adverse events. In contrast to the increased
adverse eect rates for prevention, there was no evi-
dence that amantadine recipients had increased adverse
eect rates to placebo recipients. The incidence of
adverse eects by comparison expressed as a percen-
tage of participants is shown in Table 7.
3.3.5. Comparison E oral rimantadine compared to
placebo in inuenza treatment
3.3.5.1. Ecacy. Rimantadine also signicantly shor-
tened duration of fever compared to placebo (by 1.27
days 95%CI: 0.771.77). There was a signicantly
higher number of afebrile cases 48 h after commencing
rimantadine treatment (RR=0.17; 95%CI: 0.040.74).
3.3.5.2. Adverse events. There were very little data
available for the assessment of adverse eects of
rimantadine for treatment (45 participants) (Table 8).
3.3.6. Comparison F oral amantadine compared to
oral rimantadine in inuenza treatment
3.3.6.1. Ecacy. The little data available directly
comparing amantadine and rimantadine for treat-
ment showed that the ecacy of the two drugs was
comparable, although condence intervals are verywide.
3.3.6.2. Adverse events. There were very little data
available for the assessment of adverse eects of the
direct comparison between amantadine and rimanta-
dine (33 participants).
A meta-analysis of the symptoms outcome data was
considered to further inform the assessment of the
eects of amantadine and rimantadine in a treatment
Fig. 4. Oral rimantadine compared to placebo in inuenza prevention: inuenza cases clinically dened (Expt=experimental arm, ctrl=control
arm; 95%CI=95% Condence intervals)
Table 6
Comparison B: Incidence of adverse eects expressed as a percentage of participants
Rimantadine (%) Placebo (%) No of trials N Signicant
All adverse eects 18.6 10.8 3 558 No
GI eects 9.0 2.2 2 357 Yes
Increased CNS activity (excitation) 6.5 4.3 3 652 No
Decreased CNS activity (depression) 9.6 1.0 1 228 No
Skin 0 0
V. Demicheli et al. / Vaccine 18 (2000) 9571030976
7/31/2019 Demicheli Et Al_2000_Prevention and Early Treatment of Influenza in Healthy Adults
21/74
role. When the outcome typology was tabulated it was
discovered that such a meta-analysis would be imposs-
ible (Table 9).
We resorted to using duration of fever (dened as a
temperature greater than 378C) as the only common
outcome. One drawback of this approach is the poss-
ible confounding eect of the presence of fever for a
variable length of time prior to and after entry to the
study (and hence at the moment of commencement of
treatment). However if random allocation had beenproperly carried out, this eect should disappear.
3.3.7. Comparison G oral amantadine compared to
oral aspirin in inuenza treatment
In Comparison G, based on Younkin, aspirin was
signicantly more eective than amantadine in redu-
cing the length of fever (by 0.47 days 95%CI: 0.17
0.76). This observation is based on 29 individuals.
Aspirin is well known for being a very eective anti-
pyretic and anti-inammatory drug, however it does
not inhibit viral replication and as such remains a
symptomatic remedy.
3.3.8. Comparison H inhaled amantadine compared
to placebo in inuenza treatment
In comparison H (based on Hayden's 1980 trial)
inhaled amantadine was no more ecacious than pla-
cebo in bringing down the respiratory or constitutional
symptom score (Weighted Mean Dierence 1.0
95%CI: 3.641.64 and 2.0 95%CI: 16.912.9 re-
spectively). This comparison also is based on small
numbers of participants (20). Not surprisingly, aman-
tadine caused signicantly more nasal irritation (RR
6.11 95%CI: 0.8643.3). Inhaled amantadine does
not appear to be particularly eective but has a high
incidence of local adverse eects which would make
compliance dicult.
The interpretation of Comparisons G and H is
made dicult by the small numbers involved and the
presence of single trials.
3.4. Eects of NIs
When compared to placebo, NIs are 55% (95%CIs:
2971%) eective in preventing naturally occurring
cases of laboratory conrmed inuenza and 67% eec-
tive (95%CIs: 908%) in experimental inuenza when
given intravenously. Overall NIs are 60% eective
(95%CIs: 7633%) (Fig. 5).
When the outcome is dened as cases of serologi-
cally conrmed inuenza, NIs are 74% eective
(95%CIs: 5087%) in preventing naturally occurring
inuenza and 86% eective (95%CIs: 1298%) in pre-
venting experimentally induced inuenza. Overall they
are 76% eective (95%CIs: 5587%) in preventingcases of laboratory conrmed inuenza. Both com-
pounds appear safe, but as yet no direct comparisons
have been carried out, so there can be no comment
upon their relative eects.
The adverse event prole (local nasal irritation) of
Zanamivir appears little dierent to placebo (OR 1.19
95%CIs: 0.393.62). However this may be due to
the relatively small denominator (112 individuals).
Oseltamivir appears to have a signicantly higher inci-
dence of systemic adverse eects than placebo (OR
1.68 95%CIs: 1.142.49).
Table 7
Comparison D: Incidence of adverse eects expressed as a percentage of participants
Amantadine (%) Placebo (%) No. of trials N Signicant
GI eects 13.8 13.4 3 494 No
Increased CNS activity (excitation) 3.5 5.0 2 475 No
Decreased CNS activity (depression) 56.4 65.2 3 491 Yes
Skin 0.9 0.4 2 465 No
Table 8
Comparison E: Incidence of adverse eects expressed as a percentage of participants
Rimantadine (%) Placebo (%) No of trials N Signicant
GI eects 0 0
Increased CNS activity (excitation) 28.6 28.6 1 14 No
Decreased CNS activity (depression) 0 8.3 1 31 No
Skin 0 0
V. Demicheli et al. / Vaccine 18 (2000) 9571030 977
7/31/2019 Demicheli Et Al_2000_Prevention and Early Treatment of Influenza in Healthy Adults
22/