Deconstructing the differences

A comparison of GBD 2010 and CHERG's approach to estimating the mortality burden of diarrhea, pneumonia, and their etiologies

Stephanie D. Kovacs, Edward (Kim) MULHOLLAND, Julia Bosch, Harry Campbell, Mohammad H. Forouzanfar, Ibrahim Khalil, Stephen Lim, Li Liu, Stephen N Maley, Colin D Mathers, Alastair Matheson, Ali H. Mokdad, Kate O'Brien, Umesh Parashar, Torin T Schaafsma, Duncan Steele, Stephen E Hawes, John T Grove

    Research output: Contribution to journalArticleResearchpeer-review

    2 Downloads (Pure)

    Abstract

    Background: Pneumonia and diarrhea are leading causes of death for children under five (U5). It is challenging to estimate the total number of deaths and cause-specific mortality fractions. Two major efforts, one led by the Institute for Health Metrics and Evaluation (IHME) and the other led by the World Health Organization (WHO)/Child Health Epidemiology Reference Group (CHERG) created estimates for the burden of disease due to these two syndromes, yet their estimates differed greatly for 2010.


    Methods: This paper discusses three main drivers of the differences: data sources, data processing, and covariates used for modelling. The paper discusses differences in the model assumptions for etiology-specific estimates and presents recommendations for improving future models.


    Results: IHME’s Global Burden of Disease (GBD) 2010 study estimated 6.8 million U5 deaths compared to 7.6 million U5 deaths from CHERG. The proportional differences between the pneumonia and diarrhea burden estimates from the two groups are much larger; GBD 2010 estimated 0.847 million and CHERG estimated 1.396 million due to pneumonia. Compared to CHERG, GBD 2010 used broader inclusion criteria for verbal autopsy and vital registration data. GBD 2010 and CHERG used different data processing procedures and therefore attributed the causes of neonatal death differently. The major difference in pneumonia etiologies modeling approach was the inclusion of observational study data; GBD 2010 included observational studies. CHERG relied on vaccine efficacy studies.


    Discussion: Greater transparency in modeling methods and more timely access to data sources are needed. In October 2013, the Bill & Melinda Gates Foundation (BMGF) hosted an expert meeting to examine possible approaches for better estimation. The group recommended examining the impact of data by systematically excluding sources in their models. GBD 2.0 will use a counterfactual approach for estimating mortality from pathogens due to specific etiologies to overcome bias of the methods used in GBD 2010 going forward.

    Original languageEnglish
    Article number16
    Pages (from-to)1-15
    Number of pages15
    JournalBMC Infectious Diseases
    Volume15
    Issue number1
    DOIs
    Publication statusPublished - 16 Jan 2015

    Fingerprint

    Diarrhea
    Pneumonia
    Epidemiology
    Mortality
    Cause of Death
    Information Storage and Retrieval
    Observational Studies
    Global Burden of Disease
    Child Health
    Autopsy
    Vaccines
    Health

    Cite this

    Kovacs, Stephanie D. ; MULHOLLAND, Edward (Kim) ; Bosch, Julia ; Campbell, Harry ; Forouzanfar, Mohammad H. ; Khalil, Ibrahim ; Lim, Stephen ; Liu, Li ; Maley, Stephen N ; Mathers, Colin D ; Matheson, Alastair ; Mokdad, Ali H. ; O'Brien, Kate ; Parashar, Umesh ; Schaafsma, Torin T ; Steele, Duncan ; Hawes, Stephen E ; Grove, John T. / Deconstructing the differences : A comparison of GBD 2010 and CHERG's approach to estimating the mortality burden of diarrhea, pneumonia, and their etiologies. In: BMC Infectious Diseases. 2015 ; Vol. 15, No. 1. pp. 1-15.
    @article{088c253a053d4accb953fd7b47257bbc,
    title = "Deconstructing the differences: A comparison of GBD 2010 and CHERG's approach to estimating the mortality burden of diarrhea, pneumonia, and their etiologies",
    abstract = "Background: Pneumonia and diarrhea are leading causes of death for children under five (U5). It is challenging to estimate the total number of deaths and cause-specific mortality fractions. Two major efforts, one led by the Institute for Health Metrics and Evaluation (IHME) and the other led by the World Health Organization (WHO)/Child Health Epidemiology Reference Group (CHERG) created estimates for the burden of disease due to these two syndromes, yet their estimates differed greatly for 2010.Methods: This paper discusses three main drivers of the differences: data sources, data processing, and covariates used for modelling. The paper discusses differences in the model assumptions for etiology-specific estimates and presents recommendations for improving future models.Results: IHME’s Global Burden of Disease (GBD) 2010 study estimated 6.8 million U5 deaths compared to 7.6 million U5 deaths from CHERG. The proportional differences between the pneumonia and diarrhea burden estimates from the two groups are much larger; GBD 2010 estimated 0.847 million and CHERG estimated 1.396 million due to pneumonia. Compared to CHERG, GBD 2010 used broader inclusion criteria for verbal autopsy and vital registration data. GBD 2010 and CHERG used different data processing procedures and therefore attributed the causes of neonatal death differently. The major difference in pneumonia etiologies modeling approach was the inclusion of observational study data; GBD 2010 included observational studies. CHERG relied on vaccine efficacy studies.Discussion: Greater transparency in modeling methods and more timely access to data sources are needed. In October 2013, the Bill & Melinda Gates Foundation (BMGF) hosted an expert meeting to examine possible approaches for better estimation. The group recommended examining the impact of data by systematically excluding sources in their models. GBD 2.0 will use a counterfactual approach for estimating mortality from pathogens due to specific etiologies to overcome bias of the methods used in GBD 2010 going forward.",
    keywords = "Adenoviridae, amebiasis, Article, Astrovirus, Calicivirus, Campylobacter enteritis, cause of death, childhood mortality, comparative study, covariate, cryptosporidiosis, data processing, data source, diarrhea, disease burden, Entamoeba histolytica, enteritis, Escherichia coli infection, general aspects of disease, Giardia intestinalis, Haemophilus influenzae type b, health care organization, human, Human respiratory syncytial virus, Influenza virus, lower respiratory tract infection, medical informatics, medical information, newborn death, observational study, pneumonia, risk factor, Rotavirus infection, salmonellosis, shigellosis, Streptococcus pneumoniae, verbal autopsy data, Vibrio cholerae, vital registration data, child, child health care, Diarrhea, Infantile, female, health, infant, infant mortality, male, mass screening, mortality, newborn, preschool child, procedures, regression analysis, statistical model, Child, Child Health Services, Child, Preschool, Female, Global Health, Humans, Infant, Infant Mortality, Infant, Newborn, Male, Mass Screening, Models, Statistical, Pneumonia, Regression Analysis",
    author = "Kovacs, {Stephanie D.} and MULHOLLAND, {Edward (Kim)} and Julia Bosch and Harry Campbell and Forouzanfar, {Mohammad H.} and Ibrahim Khalil and Stephen Lim and Li Liu and Maley, {Stephen N} and Mathers, {Colin D} and Alastair Matheson and Mokdad, {Ali H.} and Kate O'Brien and Umesh Parashar and Schaafsma, {Torin T} and Duncan Steele and Hawes, {Stephen E} and Grove, {John T}",
    year = "2015",
    month = "1",
    day = "16",
    doi = "10.1186/s12879-014-0728-4",
    language = "English",
    volume = "15",
    pages = "1--15",
    journal = "BMC Infectious Diseases",
    issn = "1471-2334",
    publisher = "BioMed Central",
    number = "1",

    }

    Kovacs, SD, MULHOLLAND, EK, Bosch, J, Campbell, H, Forouzanfar, MH, Khalil, I, Lim, S, Liu, L, Maley, SN, Mathers, CD, Matheson, A, Mokdad, AH, O'Brien, K, Parashar, U, Schaafsma, TT, Steele, D, Hawes, SE & Grove, JT 2015, 'Deconstructing the differences: A comparison of GBD 2010 and CHERG's approach to estimating the mortality burden of diarrhea, pneumonia, and their etiologies', BMC Infectious Diseases, vol. 15, no. 1, 16, pp. 1-15. https://doi.org/10.1186/s12879-014-0728-4

    Deconstructing the differences : A comparison of GBD 2010 and CHERG's approach to estimating the mortality burden of diarrhea, pneumonia, and their etiologies. / Kovacs, Stephanie D.; MULHOLLAND, Edward (Kim); Bosch, Julia; Campbell, Harry; Forouzanfar, Mohammad H.; Khalil, Ibrahim; Lim, Stephen; Liu, Li; Maley, Stephen N; Mathers, Colin D; Matheson, Alastair; Mokdad, Ali H.; O'Brien, Kate; Parashar, Umesh; Schaafsma, Torin T; Steele, Duncan; Hawes, Stephen E; Grove, John T.

    In: BMC Infectious Diseases, Vol. 15, No. 1, 16, 16.01.2015, p. 1-15.

    Research output: Contribution to journalArticleResearchpeer-review

    TY - JOUR

    T1 - Deconstructing the differences

    T2 - A comparison of GBD 2010 and CHERG's approach to estimating the mortality burden of diarrhea, pneumonia, and their etiologies

    AU - Kovacs, Stephanie D.

    AU - MULHOLLAND, Edward (Kim)

    AU - Bosch, Julia

    AU - Campbell, Harry

    AU - Forouzanfar, Mohammad H.

    AU - Khalil, Ibrahim

    AU - Lim, Stephen

    AU - Liu, Li

    AU - Maley, Stephen N

    AU - Mathers, Colin D

    AU - Matheson, Alastair

    AU - Mokdad, Ali H.

    AU - O'Brien, Kate

    AU - Parashar, Umesh

    AU - Schaafsma, Torin T

    AU - Steele, Duncan

    AU - Hawes, Stephen E

    AU - Grove, John T

    PY - 2015/1/16

    Y1 - 2015/1/16

    N2 - Background: Pneumonia and diarrhea are leading causes of death for children under five (U5). It is challenging to estimate the total number of deaths and cause-specific mortality fractions. Two major efforts, one led by the Institute for Health Metrics and Evaluation (IHME) and the other led by the World Health Organization (WHO)/Child Health Epidemiology Reference Group (CHERG) created estimates for the burden of disease due to these two syndromes, yet their estimates differed greatly for 2010.Methods: This paper discusses three main drivers of the differences: data sources, data processing, and covariates used for modelling. The paper discusses differences in the model assumptions for etiology-specific estimates and presents recommendations for improving future models.Results: IHME’s Global Burden of Disease (GBD) 2010 study estimated 6.8 million U5 deaths compared to 7.6 million U5 deaths from CHERG. The proportional differences between the pneumonia and diarrhea burden estimates from the two groups are much larger; GBD 2010 estimated 0.847 million and CHERG estimated 1.396 million due to pneumonia. Compared to CHERG, GBD 2010 used broader inclusion criteria for verbal autopsy and vital registration data. GBD 2010 and CHERG used different data processing procedures and therefore attributed the causes of neonatal death differently. The major difference in pneumonia etiologies modeling approach was the inclusion of observational study data; GBD 2010 included observational studies. CHERG relied on vaccine efficacy studies.Discussion: Greater transparency in modeling methods and more timely access to data sources are needed. In October 2013, the Bill & Melinda Gates Foundation (BMGF) hosted an expert meeting to examine possible approaches for better estimation. The group recommended examining the impact of data by systematically excluding sources in their models. GBD 2.0 will use a counterfactual approach for estimating mortality from pathogens due to specific etiologies to overcome bias of the methods used in GBD 2010 going forward.

    AB - Background: Pneumonia and diarrhea are leading causes of death for children under five (U5). It is challenging to estimate the total number of deaths and cause-specific mortality fractions. Two major efforts, one led by the Institute for Health Metrics and Evaluation (IHME) and the other led by the World Health Organization (WHO)/Child Health Epidemiology Reference Group (CHERG) created estimates for the burden of disease due to these two syndromes, yet their estimates differed greatly for 2010.Methods: This paper discusses three main drivers of the differences: data sources, data processing, and covariates used for modelling. The paper discusses differences in the model assumptions for etiology-specific estimates and presents recommendations for improving future models.Results: IHME’s Global Burden of Disease (GBD) 2010 study estimated 6.8 million U5 deaths compared to 7.6 million U5 deaths from CHERG. The proportional differences between the pneumonia and diarrhea burden estimates from the two groups are much larger; GBD 2010 estimated 0.847 million and CHERG estimated 1.396 million due to pneumonia. Compared to CHERG, GBD 2010 used broader inclusion criteria for verbal autopsy and vital registration data. GBD 2010 and CHERG used different data processing procedures and therefore attributed the causes of neonatal death differently. The major difference in pneumonia etiologies modeling approach was the inclusion of observational study data; GBD 2010 included observational studies. CHERG relied on vaccine efficacy studies.Discussion: Greater transparency in modeling methods and more timely access to data sources are needed. In October 2013, the Bill & Melinda Gates Foundation (BMGF) hosted an expert meeting to examine possible approaches for better estimation. The group recommended examining the impact of data by systematically excluding sources in their models. GBD 2.0 will use a counterfactual approach for estimating mortality from pathogens due to specific etiologies to overcome bias of the methods used in GBD 2010 going forward.

    KW - Adenoviridae

    KW - amebiasis

    KW - Article

    KW - Astrovirus

    KW - Calicivirus

    KW - Campylobacter enteritis

    KW - cause of death

    KW - childhood mortality

    KW - comparative study

    KW - covariate

    KW - cryptosporidiosis

    KW - data processing

    KW - data source

    KW - diarrhea

    KW - disease burden

    KW - Entamoeba histolytica

    KW - enteritis

    KW - Escherichia coli infection

    KW - general aspects of disease

    KW - Giardia intestinalis

    KW - Haemophilus influenzae type b

    KW - health care organization

    KW - human

    KW - Human respiratory syncytial virus

    KW - Influenza virus

    KW - lower respiratory tract infection

    KW - medical informatics

    KW - medical information

    KW - newborn death

    KW - observational study

    KW - pneumonia

    KW - risk factor

    KW - Rotavirus infection

    KW - salmonellosis

    KW - shigellosis

    KW - Streptococcus pneumoniae

    KW - verbal autopsy data

    KW - Vibrio cholerae

    KW - vital registration data

    KW - child

    KW - child health care

    KW - Diarrhea, Infantile

    KW - female

    KW - health

    KW - infant

    KW - infant mortality

    KW - male

    KW - mass screening

    KW - mortality

    KW - newborn

    KW - preschool child

    KW - procedures

    KW - regression analysis

    KW - statistical model

    KW - Child

    KW - Child Health Services

    KW - Child, Preschool

    KW - Female

    KW - Global Health

    KW - Humans

    KW - Infant

    KW - Infant Mortality

    KW - Infant, Newborn

    KW - Male

    KW - Mass Screening

    KW - Models, Statistical

    KW - Pneumonia

    KW - Regression Analysis

    UR - http://www.scopus.com/inward/record.url?scp=84923928841&partnerID=8YFLogxK

    U2 - 10.1186/s12879-014-0728-4

    DO - 10.1186/s12879-014-0728-4

    M3 - Article

    VL - 15

    SP - 1

    EP - 15

    JO - BMC Infectious Diseases

    JF - BMC Infectious Diseases

    SN - 1471-2334

    IS - 1

    M1 - 16

    ER -