who influenza meeting paget flu pandemic

Upload: charanmann9165

Post on 01-Jun-2018

224 views

Category:

Documents


0 download

TRANSCRIPT

  • 8/9/2019 WHO Influenza Meeting Paget Flu Pandemic

    1/30

    6/13/2012 1

    The Global Pandemic

    Mortality Burden project

    (GLaMOR)

    GLaMOR Core Team:

    NIVEL, Utrecht, The Netherlands: Madelon Kroneman, John Paget, FrancoisSchellevis, Peter Spreeuwenberg

    George Washington Universi ty, Washington DC, USA: Lone SimonsenSage Analytica, Bethesda, MD, USA: Roger Lustig, Robert Taylor

    Royal College of GPs, Birmingham, UK: Douglas Fleming

    WHO: Maria Van Kerkhove and Anthony Mounts

    NIVEL Contract with WHO-Geneva 2011

  • 8/9/2019 WHO Influenza Meeting Paget Flu Pandemic

    2/30

    GlaMOR Objective

    To estimate the global mortality burden ofthe 2009 influenza pandemic

     – Based on statistical attributions derived frommultiple regression models applied tomortality data from 2009 and prior years

    6/13/2012 2

  • 8/9/2019 WHO Influenza Meeting Paget Flu Pandemic

    3/30

    Background

    WHO’s global seasonal influenza mortalityestimate: 250,000-500,000 deaths per annum

    There were 18,209 lab-confirmed pandemic

    deaths reported to WHO (25 June 2011) – severely underestimated burden

    CDC is also working on a global 2009 estimate(using a probabilistic modeling approach)

    GLaMOR started July 2011 – report to WHO August 2012

     A WHO Pandemic Advisory Committee is

    supporting the work of the GLaMOR project

    6/13/2012 3

  • 8/9/2019 WHO Influenza Meeting Paget Flu Pandemic

    4/30

    6/13/2012

    4

    The GLaMOR Strategy Use multiple regression modeling to estimate pandemicmortality in ~20 collaborating countries

    Weekly mortality and virology data

    Use extrapolation methods to project those single-countryresults to rest of the world

    Based on GDP, pop structure, co-morbidities, access to care, more

    Strategy inspired by Murray et al (Lancet 06)Estimated the global 1918 pandemic mortality from annual mortality data

    ~20 single-

    countrypandemic

    mortality

    estimates

    Stage 1

    Global

    pandemicmortality

    estimates

    Raw

    mortality

    andvirology

    time

    series

    Stage 2

  • 8/9/2019 WHO Influenza Meeting Paget Flu Pandemic

    5/30

    Weekly Time Series for 2005-2009 or longer

    Mortality Data – 6 age groups

  • 8/9/2019 WHO Influenza Meeting Paget Flu Pandemic

    6/30

     GlaMOR Collaborating Countries

    6/13/2012 6

    Countriescollaborating withGlaMOR Stage 1

    19% of world population

  • 8/9/2019 WHO Influenza Meeting Paget Flu Pandemic

    7/30

     GlaMOR Collaborating Countries: WHO Euro area

    6/13/2012 7

    France, Germany, Spain, the Netherlands, Poland, Romania, Slovenia, United Kingdom, Israel,Denmark (all-cause only)

    Thank you for this great collaboration

    67% EU/EEA pop

    38% WHO Region pop

  • 8/9/2019 WHO Influenza Meeting Paget Flu Pandemic

    8/30

    Different types of mortality data All-cause deaths

     – More likely to be available – More sensitive (captures more deaths) – Less specific (more related to influenza)

    Respiratory & Cardiovascular deaths

     – all “I” and “J codes – Intermediate sensitivity and specificity – commonly used to estimates influenza burden

    Respiratory deaths – all “J” codes – Less sensitive (captures fewer deaths) – More specific (more related to influenza)

    Pneumonia and Influenza deaths6/13/2012 8

    Higherestimate

    Lower

    precisio

    Lowerestimate

    Higherprecision

  • 8/9/2019 WHO Influenza Meeting Paget Flu Pandemic

    9/30

    Mild impact country: Mortality Data Time SeriesCountry X:  All 4 causes,

  • 8/9/2019 WHO Influenza Meeting Paget Flu Pandemic

    10/30

    High impact country: Mortality Data Time SeriesCountry Y:  All 4 causes plus diabetes,

  • 8/9/2019 WHO Influenza Meeting Paget Flu Pandemic

    11/30

    Mild impact country - Stage 1 Model Results Age

  • 8/9/2019 WHO Influenza Meeting Paget Flu Pandemic

    12/30

    High impact country - Stage 1 Model Results Age=

  • 8/9/2019 WHO Influenza Meeting Paget Flu Pandemic

    13/30

    Mild Impact Pandemic Mortalityfor

  • 8/9/2019 WHO Influenza Meeting Paget Flu Pandemic

    14/30

    Mild Impact country 2 Pandemic Mortality for

  • 8/9/2019 WHO Influenza Meeting Paget Flu Pandemic

    15/30

    Importance of having good virological data

    15

    Virology Data – Ideally nationally representative, fully subtyped, and RSV – We used

    Influenza virology by type, sub-type (no RSV)

    FluNET-WHO data has been a fantastic resource – we have been able to concentrate our efforts on the mortalitydata and data analysis

    For Europe, we noticed some data gaps in FluNet – In those cases we asked countries to provide their national data

  • 8/9/2019 WHO Influenza Meeting Paget Flu Pandemic

    16/30

    16

    Stage 2: Four distinct strategies

    1. Survey sampling (simple projection)

    2. Matching (theory-based relations)

    3. Multiple Imputation (statistical relations)4.  Murray et al. method (Lancet, 2006)

  • 8/9/2019 WHO Influenza Meeting Paget Flu Pandemic

    17/30

    Example of Stage 2 Modeled Global Burden pattern

    6/13/2012 17

    Placeholderdarker color = higher 2009 pandemic mortality rate

  • 8/9/2019 WHO Influenza Meeting Paget Flu Pandemic

    18/30

    Summary: Stage 2

    Choice of extrapolation method – Drop survey method: too simplistic –   “Matching” and “imputation” produce different

    patterns even when using the same factors

     – Initial runs indicate Murray et al. method givesvery high estimates (as GDP based)

    Final GLaMOR estimates will be based onRespiratory deaths as a confident

    minimum burden estimateConfidence on parameter estimates better All-cause and Cardio-Respiratory estimates higher thanRespiratory in hardest hit countries, but for countrieswith low burden, respiratory [and Cardio-resp] is higher

    6/13/2012 18

  • 8/9/2019 WHO Influenza Meeting Paget Flu Pandemic

    19/30

    New initiatives for the Glamor Euro network

    WHO Euro project: country specif ic 2009 pandemicmortality estimates

     – GLaMOR will only publish Regional estimates – Publication of country-specific mortality estimates for the

    WHO Euro Region – We would like to include some more countries (?Russia?) – We would like to produce more detailed estimates (e.g. more

    age groups) – We would be able to produce more accurate estimates as the

    Stage II procedure will be based on WHO Euro countries (53

    countries). This would exclude global outliers (e.g. Americas)

    6/13/2012 19

  • 8/9/2019 WHO Influenza Meeting Paget Flu Pandemic

    20/30

    New global initiatives

    Global mortality impact for 2010 pandemic

     – Repeat analysis for 2010 and obtain more complete pictureof global pandemic impact (2009 and 2010)

    Global mortality impact of seasonal influenza – Estimate mortality impact of seasonal influenza (based on

    data for previous ten seasons) – Validation of WHO’s global seasonal mortality estimate

    of 250,000-500,000 deaths per annum

    6/13/2012 20

    A k l d t C t C ll b t

  • 8/9/2019 WHO Influenza Meeting Paget Flu Pandemic

    21/30

    6/13/2012 21

     Acknowledgments -- Country Collaborators

     AustraliaDavid Muscatello, Raina MacEntyre,Dominic.Dwyer

     ArgentinaHoracio Echenique, Vilma Savy, LuisCarlino

    BrazilWladimir Alonso, Cynthia Schuck,Fernanda Edna Araujo Moura, TerezinhaMaria de Paiva, Marcia Lopes de Carvalho

    CanadaDena Schanzer

    ChileRodrigo Fuentes, Andrea Olea,VivianaSotomayor, Fatima Marinho de Souza

    Denmark

    Jens Nielsen, Kåre MølbakGermany

    Udo Buchholz, Brunhilde Schweiger

    FranceMagali Lemaitre, Fabrice Carrat, BrunoLina, Sylvie van der Werf

    Hong KongBen Cowling, Gabriel Leung

    Israel

    Michal Bromberg, Zalman Kaufman, TamarShohat, Rita Dichtiar, Yair Amikam

    JapanNorio Sugaya, Shuichiro Hayashi, MegumiMatsuda

    MexicoHugo Lopez-Gatell-Ramirez, CeliaMercedes Alpuche Aranda, Daniel ErnestoNoyola Cherpitel

    Netherlands Adam Meijer, Kees Van den Wijngaard,Marianne van der Sande, Liselotte Van

     Asten, Mirjam Kretzschmar

    New ZealandMichael BakerRomania

    Laurentiu Zolotusca, Adriana.Pistol, FlorinPopovici, Odette Nicolae, Rodica Popescu

    A k l d t C t C ll b t

  • 8/9/2019 WHO Influenza Meeting Paget Flu Pandemic

    22/30

    6/13/2012 22

     Acknowledgments -- Country Collaborators

    PolandBogdan Wojtyniak and Daniel Rabczenko

    Singapore

    Li Wei Ang, Mark Chen, Jeffery Cutter, VernonLee, Raymond Lin

    SloveniaMaja Socan and Katarina Prosenc

    Spain Amparo Larrauri Cámara, Salvador de Mateo and

    colleaguesSouth KoreaUn-Yeong Go, Yeonkyeng Lee

    South AfricaCheryl Cohen

    United Kingdom

    Richard Peabody, Helen GreenUSA

    David Shay, Joseph Breese, Cecile Viboud

  • 8/9/2019 WHO Influenza Meeting Paget Flu Pandemic

    23/30

  • 8/9/2019 WHO Influenza Meeting Paget Flu Pandemic

    24/30Thank you for your attention

  • 8/9/2019 WHO Influenza Meeting Paget Flu Pandemic

    25/30

    6/13/2012 25

    Extras

  • 8/9/2019 WHO Influenza Meeting Paget Flu Pandemic

    26/30

    6/13/2012 26

    Lab-confirmed H1N1 pandemic deaths reported to WHO

    Pandemic (H1N1) 2009 - update 106 (25 June 2011): http://www.who.int/csr/don/2010_06_25/en/index.html

  • 8/9/2019 WHO Influenza Meeting Paget Flu Pandemic

    27/30

    6/13/2012 27

    Pros and Cons of

    GLaMOR approach

    Pros

     – Based on actual mortality data

     – Comparable between countries (same model, outcomes)

     – Comparable to historic pandemic burden estimates

     – Comparable to seasonal burden estimates

    Cons

     – Need up-to-date weekly mortality data

     – Need up-to-date weekly virology data

    Especially for tropics, sub-tropics, and for 2010-11 due to H1N1 co-circulation with A/H3 and B

    RSV lab data (or proxy) needed to control for RSV mortality burden

     – Must project an ~25 country convenience sample to 170 countriesaround the world.

    Model Form

  • 8/9/2019 WHO Influenza Meeting Paget Flu Pandemic

    28/30

    6/13/2012 28

    Model FormMultiple regression (additive model)

    Outcomet=β1t + β2t

    2+ β3t3 +β4-5sin(2Πt/w)+ β6-7cos(2Πt/w) 

    + β7FluAt + β8FluBt + (β10RSVt)

    FluA dummy variables,

    For each season This βxFluAt for 2009yields the pandemicinfluenza attribution

    Pandemic Excess Mortality=

    =Σ βxFluAt for pandemic period, Apr1 to Dec31 2009

    Dropped RSVTerm due to lack ofVirology data

    sin, cos terms for ½and 1 year cycles

    Virology data= numbers (not %)

    Deaths/100,000 popt = running week number  

    Dropped log linkSAS Proc Reg

    13/06/2012 28

    Stage 1 Model Fit

  • 8/9/2019 WHO Influenza Meeting Paget Flu Pandemic

    29/30

    Stage 1 Model Fitand contribution of influenza terms in model (lift)Lift Table, expanded age groups, all outcomes

    Model also fi ts well for all age groups, except for diabetes outcomes,

    with a strong lift” for respiratory outcomes in age groups 5-6413/06/2012 29

  • 8/9/2019 WHO Influenza Meeting Paget Flu Pandemic

    30/30

    Exampleof model

    output

     Agegrp:Under 65 years

    Outcome:Respiratory

    Model

     Adj R2=.85

    Pandemic Flu Termsignificant (p