Vaccine 24 (2006) 64686475

Influenza-related mortality in the Italiacci

EmanA. MiversityethesdaPublic

ase, NIH

ne 200006

Abstract

We investigated trends in influenza-related mortality among the elderly population in Italy associated with increased vaccination coverage.Using Italian vital statistics data, we studied monthly death rates for pneumonia and influenza and all-cause for persons 65 years ofage by 5-yeexcess mortfor populatiinfluenza sudistribution(possibly duchallenges cvaccination 2006 Else

Keywords: In

1. Introdu

The incsubstantialditions [1,2years and orelated deaeach decad

Influenzfirmed, andto co-morboccur after

CorresponE-mail ad

0264-410X/$doi:10.1016/jar age groups for 19702001. Using a classic seasonal regression modelling approach, we estimated the age-specific seasonalality rates among Italian elderly as a measure of influenza-related deaths. We studied trends in excess mortality after adjustingon aging and analyzing separately seasons dominated by the severe A/H3N2 subtype and those dominated by other circulatingbtypes. After the late 1980s, no decline in age-adjusted excess mortality was associated with increasing influenza vaccinationprimarily targeted for the elderly. These findings suggest that either the vaccine failed to protect the elderly against mortalitye to immune senescence), and/or the vaccination efforts did not adequately target the frailest elderly. As in the US, our studyurrent strategies to best protect the elderly against mortality, warranting the need for better controlled trials with alternativestrategies.vier Ltd. All rights reserved.

fluenza; Vaccination; Mortality

ction

idence of severe outcomes of influenza increasesly with age and for individuals with high-risk con-]. During seasonal epidemics persons aged 65lder account for more than 90% of all influenza-ths with an exponential increase in death rates fores past 65 years [3,4].a diagnoses are generally not laboratory con-

deaths related to influenza are often attributedid conditions or to secondary complications thatthe influenza virus infection [5]. For these rea-

ding author. Tel.: +39 06 4990 4277; fax: +39 06 44232444.dress: caterina.rizzo@iss.it (C. Rizzo).

sons, influenza-related mortality is traditionally quantifiedindirectly using statistical methods that estimate the sea-sonal increases in pneumonia and influenza (P&I) or all-cause(AC) mortality above a baseline of expected mortality, whichoccurs concurrently with influenza virus circulation [611].Since the 1968 influenza A(H3N2) Hong Kong pandemic,the seasonal influenza-related mortality burden has fluctuatedwidely, depending on which one of the circulating virus sub-types (influenza A(H3N2), A(H1N1) and B) dominate. Thehighest mortality rates typically occur in seasons dominatedby A(H3N2) viruses [12].

During the last three decades, the life expectancy and theproportion of elderly in the Italian population has increasedmore than in other European countries In particular, the popu-lation size of elderly >85 years of age has doubled from 1990

see front matter 2006 Elsevier Ltd. All rights reserved..vaccine.2006.06.052associated with increasing vaCaterina Rizzo a,, Cecile Viboud b,

Lone Simonsen d, Marka Department of Pharmaco-Biology, Un

b Fogarty International Centre, NIH, Bc Department of Physiopathology Experimental Medicine and

d National Institute of Allergy and Infectious DiseReceived 5 May 2006; received in revised form 19 Ju

Available online 7 July 2n elderly: No declinenation coverageuele Montomoli c,iller b

of Bari, Italy, MD, USAHealth, University of Siena, Italy, Bethesda, MD, USA

6; accepted 20 June 2006

C. Rizzo et al. / Vaccine 24 (2006) 64686475 6469

to 2001 [13]. With this ageing of the population one mightexpect an increase in influenza related mortality, due to theexponential risk of influenza-related death with age amongelderly [3,4been recomthose with cconditions)ing long-tethose whoand may trage amongfrom 19802001 [15,1

There isinfluenza vand nationcohort studvaccinatedtion in all wever, such cmay lead toRecent cohfound simicinated andperiods, indtection frothat adjustmcodes andquate [21].influenza-rof mortalit[12]. The lelderly at threview of ithat the anlower than

To evaluto the increand confir[12], we intality in Itain virus suuse a meth[12].

2. Method

2.1. Morta

The moand all-caucertificates(ISTAT) frofrom undertional Clas

470474 from ICD-8th (19701979) and 480486 and 487from ICD-9th revisions (19792001).

For each year, we generated summary data sets of thehly nu-year aor 8ed moe grou

in 19e obtanal Cnnuallated, and

Inue

om 19on dby th

ina Ds andopulane doccinatroup,basedhom vn is reger pears ofsentinated t1980butedr mor

ne covnstitu

Virolo

r eachal Wesubtynza s

ned th3N2),from t26] ane cons

ountein th

es prenated]. In Italy, since 1980 influenza vaccination hasmended for people 65 years of age and older,hronic diseases (respiratory, cardiovascular, renal, children under 12 years of age who are receiv-rm anti-inflammatory treatment with aspirin, andhave frequent contact with these high risk groupsansmit influenza to them [14]. Vaccination cover-the elderly population in Italy increased slowly

to 1987, then rose sharply to 60% from 1988 to6] (Table 1).

considerable discrepancy in measurements ofaccine benefits in observational cohort studiesal trend studies. A large body of observationalies comparing mortality in vaccinated and non-elderly have consistently reported a 50% reduc-inter deaths in vaccinated elderly [17,18]. How-ohort studies are prone to self-selection bias thatsubstantial overestimation of mortality benefits.

ort studies set in the UK [19] and the US [20]lar reductions in all-cause mortality between vac-

unvaccinated elderly outside influenza epidemicicating a vaccination receipt bias rather than pro-

m influenza-related mortality. It was also foundent procedures using broad disease classification

typically used for cohort studies were not ade-Moreover, a recent study of long-term trends in

elated mortality among US elderly reported a lacky benefits associated with the vaccination effortsack of observed vaccine mortality benefits in thee population level is further supported by a recent

nfluenza vaccine immunogenicity, demonstratingtibody response in the elderly was considerablythat of younger adults [22].ate the mortality benefits that may be attributedasing vaccination coverage in the Italian elderly

m or infirm the results of the US trend studyvestigate trends in influenza-related excess mor-ly, adjusting for population aging and changes

btype circulation. For comparison purposes, weodological approach very similar to the US study

s

lity and population data

nthly numbers of pneumonia and influenza (P&I)se (AC) deaths in Italy were obtained from deathcollected by the Italian National Census Bureaum 1970 through 2001. We identified P&I deathslying cause of death statistics from the Interna-sification of Diseases (ICD), codes 480486 and

montby 5and fstudian ag(6%

WNatiothe acalcugroup

2.2.years

FrbasedtinelyDr. Dmentthe pvacciof vaage gcinefor wnatioyoun12 yerepreestimfromdistri

Fovaccinon-i

2.3.

Foannu

viralinflueobtai(A(Hdata([25,

Wit acctypedvirusdomimbers of influenza, P&I and AC deaths, stratifiedge intervals (6569, 7074, 7579 and 8084),5 years old. As a non-historic control, we alsortality trends in a younger age group 4564 years,p associated with a very low vaccination coverage992001).

ined annual population estimates from the Italianensus Bureau from 1970 to 2001, and calculatednumber of elderly in each age group. We then

monthly mortality rates per 100,000 for each agestandardized these to 30.4-day months.

nza vaccination coverage in people over 65

80 to 1999 our estimates of vaccine coverage wereata on vaccine distribution and use collected rou-e Ministry of Health (personal communication:e Stefano Caraffa) from the local health depart-general practitioners who administer vaccines totion. These data represented the total number ofses distributed, and provide an indirect measureion coverage. Although the data is not specific bywe assumed a proportionate distribution of vac-on the proportion of the total population groupsaccine is recommended. In Italy, influenza vacci-commended [14] for those over 65 years of age,rsons with chronic conditions, and children underage on long-term aspirin therapy, the latter two

g 8% of the total population [13]. Therefore, wehat the overall number of doses distributed in Italy1999 was a good proxy for the number of dosesin the elderly population.e recent years, 19992001, age-specific influenzaerage was estimated from the distribution rates in

tionalized elderly 65 years of age [23].

gical surveillance

influenza season from 1970 to 1980, we reviewedekly Epidemiological Report listing influenzapes identified in Italian laboratories during eacheason [24]. For seasons from 1980 to 2001, wee proportion of each circulating influenza subtypeA(H1N1), and B), using laboratory surveillance

he Italian National Influenza Centre of the WHOd Dr. Donatelli, personal communication).idered an influenza subtype to be dominant whend for at least 50% of all isolates that were sub-at season. Of the 31 seasons studied, A(H3N2)dominated in 20, the remaining 11 seasons wereby A(H1N1) or influenza B viruses (Table 1).

6470C.Rizzo

etal./V

accine

24(2006)64686475

Table 1Dominant influenza virus subtype, influenza vaccination coverage, total winter and seasonal excess all-cause (AC) death rate in person aged 65 years for 31 influenza seasons, 19702001 in ItalyInfluenz seasons Dominant

viral subtypePercentage ofdominant virus

Italian elderly 65 years of ageVaccinationcoverage (%)

Total Italian winter mortality(DecemberMarch)

Excess AC death(regression model)

Excess AC deathrate/100000

Percentage Excess ACmortality/all winter deaths

1970/1971 B 131600 5600 95.1 4.31971/1972 A(H3N2) 136400 8700 141.4 6.41972/1973 A(H3N2) 150800 21400 341.1 14.21973/1974 A(H3N2) 138800 1100 17.9 0.81974/1975 A(H3N2) 160300 24800 377.0 15.51975/1976 A(H3N2) 152500 10300 153.3 6.81976/1977 A(H3N2) 155800 9900 144.7 6.41977/1978 A(H1N1) 150400 2400 34.3 1.61978/1979 A(H1N1) 148300 0 0 0.01979/1980 A(H3N2) 160000 8400 113.8 5.21980/1981 A(H3N2) 55 5 165600 13900 185.5 8.41981/1982 A(H3N2) 75 5 148000 0 0 0.01982/1983 A(H3N2) 88 6 163500 17200 232.6 10.51983/1984 A(H1N1) 93 6 151100 2900 39.4 1.91984/1985 A(H3N2) 89 7 155300 0 0 0.01985/1986 A(H3N2) 61 7 162900 13000 173.8 8.01986/1987 A(H1N1) 91 8 149800 0 0 0.01987/1988 B 61 10 147900 2400 31.1 1.71988/1989 A(H1N1) 97 14 151800 2700 33.0 1.81989/1990 A(H3N2) 96 18 165600 12500 150.8 7.61990/1991 B 74 22 158400 0 0 0.01991/1992 A(H3N2) 99 26 161800 4500 51.5 2.81992/1993 B 60 29 164200 6100 67.8 3.71993/1994 A(H3N2) 85 32 162800 4900 53.6 3.01994/1995 A(H3N2) 63 35 164500 4100 44.1 2.51995/1996 A(H3N2) 62 38 166300 2700 28.0 1.61996/1997 A(H3N2) 87 41 175400 9500 96.5 5.41997/1998 A(H3N2) 93 49 174800 10800 107.9 6.21998/1999 Ba 58 55 189900 21400 210.1 11.31999/2000 A(H3N2) 99 61 183200 14500 140.2 7.92000/2001 A(H1N1) 99 61 162100 0 0 0.0Average for all seasons 158400 7600 988 4.8

a Influenza B viruses and influenza A(H3N2) viruses co-dominated.

C. Rizzo et al. / Vaccine 24 (2006) 64686475 6471

2.4. Statistical method

2.4.1. Excess mortality modelTo estimate age-specific excess P&I and AC mortality rate

for 31 influenza seasons, 19702001, we applied a Serfling-type regression model to monthly data [8], as described in theUS study of trends in influenza-related mortality [12]. Briefly,we applied a seasonal regression model to a de-trended seriesof death rates, excluding values for DecemberApril, to quan-tify the expected mortality in the absence of influenza activ-ity. Monthly excess mortality rates were calculated as theobserved minus expected mortality rate for all influenza epi-demic months. Epidemic months were identified for each sea-son based on those deaths where influenza was mentionedas the underlying cause of deaths (ICD-8 code 470474 andICD-9 code 487). Seasonal excess mortality was estimated asthe sum of monthly excess mortality, after back adjusting forthe true month length and time trend. We achieved an excel-lent fit for all age groups. All model terms included were sta-tistically significant (p < 0.0001), but additional terms werenot (p > 0.05).

2.4.2. Adjustment of seasonal excess mortality rates forpopulation aging

We applied age-specific seasonal death rates by 5-year agegroup to the Italian population in 1975, considered a refer-ence populby summinest (65+, 6populationold and ovegroup we udardization

2.4.3. Statistical analysis of trendsTo assess changes in excess mortality rates before and

during the period of increasing vaccine coverage, we fit lin-ear regression models for logarithmic (log 10) excess mor-tality rate on year, age, and ageyear interaction for twoperiods, 19701986 (a period with no or low vaccine cov-erage) and 19872001 (a period with sharply rising vaccinecoverage).

3. Results

3.1. Mortality impact of inuenza among Italian elderly

Of the 31 influenza seasons studied in Italy, we identi-fied a seasonal average of 99 excess AC deaths per 100,000elderly 65 years (Table 1). Seasonal excess AC influenzadeaths never represented more than 15% of the approximately160,000 deaths that occur among the elderly each winter.Six of the 31 seasons (1978/1979, 1981/1982, 1984/1985,1986/1987, 1990/1991 and 2000/2001) were not associatedwith measurable excess mortality. The most severe seasonswere characterized by a predominance of A(H3N2) viruses.Overall, the average P&I death rates for season dominatedby A(H3N2) influenza viruses was 4.2-fold greater than thatof the 11 s

es (Tae 3-yortali

asingb). Mbutionfor ex

Fig. 1. Season sons 19death rates are d rates sor A/H1N1). T A/H3Nto color in thiation. Age-standardized rates were then obtainedg excess deaths in combined age groups of inter-574, 7584) and dividing the sum by the 1975of that combined age group. Because the 85 yearsr age class data were not available by 5-year agesed it as an open-ended category without stan-.

virusTh

AC mincrea anddistritime;

al excess mortality rates among elderly 65 years of age, for individual seaindicated. (a and b) Unadjusted rates (red square) and (c and d) age-adjustehe red curve represents a 3-year moving average of all seasons (a and b) and

s figure legend, the reader is referred to the web version of the article.)easons dominated by A(H1N1) and B influenzable 2).ear moving averages of unadjusted excess P&I andty excess rates in persons aged 65 years shows atrend in the last decades of the study (Fig. 1: panelsoreover there was a pronounced change in the ageof influenza-related deaths within the elderly over

ample, elderly >85 years of age accounted for 26%

702001, Italy. P&I (left panels) and AC (right panels) excesstratified by virus subtypes (red dots: A/H3N2; blue squares: B2-only seasons (c and d). (For interpretation of the references

6472 C. Rizzo et al. / Vaccine 24 (2006) 64686475

Table 2Mean seasonal excess death rates per 100,000 from pneumonia and influenza (P&I) and all-cause (AC) in Italy 19702001 by age and dominant influenzasubtype (crude and age adjusted)Subtype Excess mortality (1000001)

4564 yearsExcess mortality (1000001)>65 years (crude)

Excess mortality (1000001)>65 years (age-adjusteda)

P&I AC P&I AC P&I AC

A(H3N2) 1.04 7.53 19.37 127.69 18.34 118.81A(H1N1)/B 0.15 2.09 4.61 46.44 4.00 40.77Total 0.72 5.60 14.13 98.86 13.25 91.11

a Excess death rates are age-adjusted to the population of Italy in 1975 by 5-year age groups.

of excess AC deaths in the early 1970s but had reached 34%in the 1990s.

For the 20 influenza seasons dominated by A(H3N2)viruses, we estimated average excess P&I and AC deathrates of 1965 yearsthan for thP&I and Aamong eldseasonal Pdecade studof the studadjusted esmate durinwe founddominated1970s, thethe A/H3Nmarked destrong andsubtype.

For theand/or B viwere low csons (Tableentire study

3.2. Trendinuenza A

The 3-yAC mortalisharply fro

0.05).

s in inuenza-related mortality for 20(H3N2) seasons

ear moving averages of unadjusted excess P&I andty excess rates in persons aged 65 years declinedm 1970 to mid-1980s, when there was an estimated

mid-A

the mpletemortafterand Apane

Thall eperioadjusand60%but rmostwas

loweelderAC mthe la

Todurinexploing,usedpopuinterAC m

ess mortality rates in Italian elderly (6584a years) for 20 influenza A(H3Ncreasing vaccine useP&I

%Annualchange

%Total change duringthe time period

p-Valu

use (

C. Rizzo et al. / Vaccine 24 (2006) 64686475 6473

Fig. 2. Age-sseasonal exced). Unadjusteand blue squa65, and the recolor in this fi

significantexcess mor

evidence ofcoverage ro(p = 0.75) m

3.3. Sensit

We conresults werincluded inticular, 199curiously winfluenza Ba high proreported inseason ma

son as a m

ses, it onlynot detectpecific seasonal excess mortality rates among elderly 65 years of age, for indss P&I (left panels) and all-cause (AC) (right panels) mortality rates (per 100,000)d excess death rates (per 100,000) are shown for people 85 years (e and f). Red dores for seasons dominated by influenza A(H1N1) and/or B viruses. The black solidd line represents a 3-year moving average of all seasons (a and b) and A/H3N2-ogure legend, the reader is referred to the web version of the article.)

declining trend over time for both P&I and ACtality rates (p 0.002). By contrast, we found noany trend during the second period when vaccinese (19872001), either for P&I (p = 0.36) or ACortality.

ivity analysis on trends ndings

ducted additional analyses to assess whether oure sensitive to the choice of the subset of seasonsthe trends analysis. One influenza season in par-81999, had high excess mortality impact, butas defined as an influenza B season (in whichrepresented 58% of all isolates). Interestingly,

portion of co-circulating A(H3N2) viruses wasthat same season (42%), suggesting that this

y be misclassified. When we included this sea-isclassified A(H3N2) season in the trend analy-marginally affected our findings: we still could

any significant trend in influenza-related mortal-

ity rates incoverage.

4. Discuss

We haveity in the Itsought to ecination prlate 1970s,than 50% acine distribthe year 20among persactual 60years [23].

After caing by domfrequencyividual seasons 19702001, Italy. Age-adjusted estimates ofamong person 6574 (a and b) and 7584 years of age (c andts indicate seasons dominated by influenza A(H3N2) viruses,

line shows influenza vaccination coverage rates in people overnly seasons (c and d). (For interpretation of the references to

the recent period of rapidly increasing vaccine

ion

studied the influenza associated excess mortal-alian elderly population, from 1970 to 2001, andvaluate the mortality benefits of the influenza vac-ogram. Although the program was initiated in thewe estimate that it only achieved coverage greatermong the elderly by the late 1990s, based on vac-ution data. The Italian National Vaccines Plan for00 [27] had established a goal of 75% coverageons 65 years or older, considerably higher than the% reported from the Ministry of Health in recent

refully adjusting for population aging and stratify-inant influenza sub-type to account for increasingof circulation of the virulent A(H3N2) viruses in

6474 C. Rizzo et al. / Vaccine 24 (2006) 64686475

recent years, we found no evidence of reduction in influenza-related mortality in the last 15 years, despite the concomi-tant increase of influenza vaccination coverage from 10%to 60%.attributableaverage, anfindings arusing the saall winter d[12]. Thesefrom cohocan reduce[17,18]. Thnation covewas not assin line withtrend studiepattern obslower vacc

In the yobserved apeople ageacquisitionfirst decadeas vaccinafewer influcould be prdeath rategroups (anBy contrasproportionpeople agedeath of 31

4.1. Cavea

In the trthe elderlyto die of insons such achronic disbenefits ofof this effecwhere sumunrelated tgroups, sum1980 (by gests that ththan the eldadjusted foeven less reyears of inc

Anotherinfluenza vfrom the totitioners an

consistent with limited regional studies that have surveyedvaccination rates in 1995 and 1999 [15,16,29]. Additionally,our data are in agreement with other studies of influenza vac-

sales ine usethe 2

e 1).pe anly has[30].

fficienram, ifstly,

ly adde appaxampbeenmaskee res

in a simgies fiated

is agene effpossibsystemst eldIndeert studccinereduc

introdnd Italhe verbly duperti

4 of al). The

te aboenza ancludiple orf anti

l adjuis too

ely, ingh effty. Vahool cailest

rence

urrayorld: G26976Furthermore, the proportion of all winter deathsto influenza as AC excess mortality was 4.9% on

d always less than 15% for any season studied. Oure consistent with those of a recent US trend studyme methodology as ours, where the proportion ofeaths attributable to influenza was 5% on averagefindings cannot be reconciled with conclusions

rt studies which state that influenza vaccinationall winter mortality by 50% in those vaccinatede US trend study found that the increase in vacci-rage (from 1520% before 1980 to 65% in 2001)ociated with a decline in influenza death rates, alsoour results. Moreover, in both the US and Italians, findings were confirmed by a similar mortalityerved for persons aged 4564, an age group withination coverage.ears immediately after the 1968 pandemic, wesharp decline in influenza-related deaths among

d 6574 and 7584, that could be explained by theof natural immunity to these viruses during theof A(H3N2) emergence [28]. By the late 1980s,

tion coverage increased, there were relativelyenza-related deaths in the younger elderly thatevented by vaccination, i.e. the 6574 (an averageof 17 per 100,000 per season) and 7584 ageaverage death rate of 83 per 100,000 per season).t, the very elderly accounted for an increasingof influenza deaths over time; in particular,

d 85 and over have a risk of influenza-related4 per 100,000 per season.

ts

end analysis, we considered the possibility thatpopulation in the 1990s were more frail and likelyfluenza than their age peers of the 1980s for rea-s increased medication use and living longer witheases. Such a cohort effect could mask mortalitythe vaccine. However, we rejected the possibilityt by comparing summer mortality rates over time,mer mortality is good control of cohort effectso influenza. We found that for all elderly age

mer AC mortality rates declined steadily after1% every year, p < 0.001), which strongly sug-e current elderly population is actually less frailerly population of earlier years. If we had furtherr such cohort effects, it would have resulted induction in influenza-related mortality during thereasing vaccination coverage.caveat relates to possible overestimation of the

accination coverage before 1999, as it was derivedtal number of doses distributed by general prac-d local health departments. Yet, our data are

cinevaccileling(TablEuroelderyearsbe suprog

Lapleteto thFor ehavehave

Thingsstrateassoc

in thvacciOneof afraile[20].coho

Vaing atheirUS athat tpossiciallyfor 3/studydebainflugies imultuse o

nove

but itnativthroumunior sc

the fr

Refe

[1] Mw

1n Italy, which documented a 2.8-fold increase inbetween 1990 and 2000 [30,31], closely paral-

.7-fold increase that we report in the same periodHence in Italy as in several other countries ind North America, vaccine coverage among theincreased from

C. Rizzo et al. / Vaccine 24 (2006) 64686475 6475

[2] Sprenger MJ, Mulder PG, Beyer WE, Van SR, Masurel N. Impactof influenza on mortality in relation to age and underlying disease,19671989. Int J Epidemiol 1993;22(2):33440.

[3] Istituto Nazionale di statistica. Cause di Morte, Anno 2000. ISTAT,editor. Annuario 16, 2004. ISTAT.

[4] Simonsen L, Clarke MJ, Schonberger LB, Arden NH, Cox NJ, FukudaK. Pandemic versus epidemic influenza mortality: a pattern of changingage distribution. J Infect Dis 1998;178(1):5360.

[5] Simonsen L. The global impact of influenza on morbidity and mortality.Vaccine 1999;17(Suppl 1):S3S10.

[6] Noble GR. Epidemiological and clinical aspects of influenza. In: Beare,editor. Basic and applied influenza research. Boca Raton, FL: CRCPress; 1982.

[7] Farr W. Vital statistics: a memorial volume of selections from the reportsand writings of William Farr. London: Office of the Sanitary Institute;1885.

[8] Serflingpneumon

[9] Lui KJ, KUnited S1987;77(

[10] ReichertMA. Infl195919

[11] CrombiemonthlyJ Epidem

[12] SimonseMiller Mity in th26572.

[13] Istituto Ndi vita e

[14] Ministrystagione

[15] SammarVaccine2004;16(

[16] Pregliascamong t12731.

[17] Gross PAefficacyreview o

[18] Vu T, Faof influecommun

[19] MangtanAJ. A copeople, pdatabase

[20] Jacksonof bias inEpidemi

[21] Jackson LA, Nelson JC, Benson P, et al. Functional status is a con-founder of the association of influenza vaccine and risk of all causemortality in seniors. Int J Epidemiol 2006;35(2):34552.

[22] Goodwin K, Viboud C, Simonsen L. Antibody response to influenzavaccination in the elderly: a quantitative review. Vaccine 2006;24(8):115969.

[23] Ministry of Health. Influenza Vaccination Coverage. http://www min-isterosalute it/promozione/malattie/malattie jsp 2005.

[24] [Weekly epidemiological report; Archives 19261995]. Scalpel(Brux) 1980Available from: URL: http://www.who.int/wer/archives/en/.

[25] Mancini G, rangio-Ruiz G, Donatelli I, Rozera C, Butto S. Diffusionof influenza viruses in Italy in years 1982-83-84. Laboratory investiga-tions at the National Influenza Centre of WHO. Ann Ist Super Sanita1986;22(2):71721.

[26] Mancini G, rangio-Ruiz G, Campitelli L, et al. Surveillance of influenzaand B

988;4(4onferenazionana, volimonseonsequeew Yor004. p.Argen, GrecoellItaliedson Dnfect Dmbroscpdate tooronzyeyand

iveness001;75(ohen Jlderly. Simonseesearchuckried

or influeeitel Waccine aed 200

odda Aants: e001;19(odda Aaper JBaccinesallorans for006;311RE. Methods for current statistical analysis of excessia-influenza deaths. Public Health Rep 1963;78:494506.endal AP. Impact of influenza epidemics on mortality in the

tates from October 1972 to May 1985. Am J Public Health6):7126.TA, Simonsen L, Sharma A, Pardo SA, Fedson DS, Miller

uenza and the winter increase in mortality in the United States99. Am J Epidemiol 2004;160(5):492502.DL, Fleming DM, Cross KW, Lancashire RJ. Concurrence ofvariations of mortality related to underlying cause in Europe.iol Community Health 1995;49(4):3738.

n L, Reichert TA, Viboud C, Blackwelder WC, Taylor RJ,A. Impact of influenza vaccination on seasonal mortal-e US elderly population. Arch Intern Med 2005;165(3):

azionale di statistica. Indagine Multiscopo sulle famiglie: stilicondizioni di salute. ISTAT, editor [5], 2005. ISTAT.of Health. Profilassi antinfluenzale. Raccomandazioni per la1989-1990. Circolare Ministeriale no. 28; 1989.co S, Ciofi degli Atti ML, Binkin N, DArgenio P, Bella A.coverage survey of the elderly in 11 Italian regions. Ann Ig6):7018.o F, Sodano L, Mensi C, et al. Influenza vaccinationhe elderly in Italy. Bull World Health Organ 1999;77(2):

, Hermogenes AW, Sacks HS, Lau J, Levandowski RA. Theof influenza vaccine in elderly persons. A meta-analysis andf the literature. Ann Intern Med 1995;123(7):51827.rish S, Jenkins M, Kelly H. A meta-analysis of effectivenessnza vaccine in persons aged 65 years and over living in theity. Vaccine 2002;20(1314):18316.i P, Cumberland P, Hodgson CR, Roberts JA, Cutts FT, Hallhort study of the effectiveness of influenza vaccine in oldererformed using the United Kingdom general practice research. J Infect Dis 2004;190(1):110.LA, Jackson ML, Nelson JC, Neuzil KM, Weiss NS. Evidence

estimates of influenza vaccine effectiveness in seniors. Int Jol 2006;35(2):33744.

A1

[27] CNia

[28] Sc

N2

[29] DGd

[30] FI

[31] Au

[32] GWs

2[33] C

e

[34] SR

[35] Hf

[36] Kv

M[37] P

v

2[38] P

Kv

[39] Hie2viruses in Italy between 1984 and 1987. Eur J Epidemiol):44550.za Permanente per i rapporti tra lo Stato lRelPA. Piano

le Vaccini 19992000. Gazzetta Ufficiale della Repubblica Ital-. 144, 1999. p. 544.n L, Reichert TA, Miller MA. The virtues of antigenic sin:nces of pandemic recycling on influenza-associated mortality.k, NY: Elsevier Science B.B. International Congress Series;7914.io P, Manfredi Selvaggi MT, Romano A, Ponzio G, PenninoD. La Vaccinazione Antinfluenzale tra gli Anziani in tre Citta`

a Meridionale. LIgiene Moderna 1995;103:20922.S. Pandemic influenza and the global vaccine supply. Clin

is 2003;36(12):155261.h F, Fedson DS. Influenza vaccination in 29 countries. An1997. Pharmacoeconomics 2003;16(S1):4754.JJ, Fulbright JW, Crowson CS, Poland GA, OFallon WM,

CM. Value of immunological markers in predicting respon-to influenza vaccination in elderly individuals. J Virol

24):121827., Influenza. Study questions the benefits of vaccinating thecience 2005;307(5712):1026.

n L, Viboud C, Blackwelder WC, Taylor R, Miller M.ers defend influenza vaccine study. Infect Dis News 2005.e A, Bungener L, Stegmann T, et al. The virosome conceptnza vaccines. Vaccine 2005;23(Suppl 1):S2638.A, Atmar RL, Cate TR, et al. Safety of high doses of influenzand effect on antibody responses in elderly persons. Arch Intern6;166(10):11217.. The adjuvanted influenza vaccines with novel adju-

xperience with the MF59-adjuvanted vaccine. Vaccine1719):267380., Del Giudice G. MF59 adjuvant emulsion. In: Levine MM,, Rappuoli R, Liu MA, Good MF, editors. New generation

. 3rd ed. New York: Marcel Dekker; 2005. p. 22535.ME, Longini Jr IM. Public health. Community stud-vaccinating schoolchildren against influenza. Science

(5761):6156.

Influenza-related mortality in the Italian elderly: No decline associated with increasing vaccination coverageIntroductionMethodsMortality and population dataInfluenza vaccination coverage in people over 65 yearsVirological surveillanceStatistical methodExcess mortality modelAdjustment of seasonal excess mortality rates for population agingStatistical analysis of trends

ResultsMortality impact of influenza among Italian elderlyTrends in influenza-related mortality for 20 influenza A(H3N2) seasonsSensitivity analysis on trends findings

DiscussionCaveats

References