Global Burden of Disease Study: Difference between revisions

Latest revision as of 15:17, 20 December 2017

[show] This page is a study. The page identifier is Op_en4685
Moderator:Nobody (see all) Click here to sign up.

Upload data {{#opasnet_base_link:Op_en4685}}

Study Overview

The new Global Burden of Diseases, Injuries, and Risk Factors Study (the GBD 2010 Study), which commenced in the spring of 2007, is the first major effort since the original GBD 1990 Study to carry out a complete systematic assessment of the data on all diseases and injuries, and produce comprehensive and comparable estimates of the burden of diseases, injuries and risk factors for two time periods, 1990 and 2005. By Spring 2011 the project will produce a final set of estimates.

The GBD 2010 Study brings together a community of experts and leaders in epidemiology and other areas of public health research from around the world to measure current levels and recent trends in all major diseases, injuries, and risk factors, and to produce new and comprehensive sets of estimates and easy-to-use tools for research and teaching. It is led by a consortium including Harvard University, the Institute for Health Metrics and Evaluation at the University of Washington, Johns Hopkins University, the University of Queensland, and the World Health Organization (WHO). This ambitious effort will be conducted systematically and transparently; both its methods and results will be made available to the public.

The GBD assesses the magnitude of health problems using standard units of measurement, such as disability-adjusted life years (DALYs). This study feature allows for lives in every part of the world to be valued equally and creates a common unit of currency for making decisions about the costs and benefits of various health interventions.

Finally, the study will focus from the outset on education, training, and transparency, incorporating features like an interactive website where experts can post information and actively discuss the study process. Broadening the global community's engagement with health metrics will be a focal point of the project.

Materials and methods

Data

2016 Study

See notes for importing the GBD2016 data to Wikidata [1].

USE CITATION URL: http://ghdx.healthdata.org/gbd-results-tool

IHME-GBD-2016 Data and metadata: Global deaths and DALYs for each disease; and Country-specific total deaths and DALYs. Also all classifications used.

Permanent link to the global data per disease fetched on 14.12.2017 for visualisations and uploading to Wikidata. [2] [3]
Permanent link to country-specific data for total deaths and DALYs fetched on 18.12.2017 for uploading to Wikidata. [4] [5]
Codebook contains explanations and hierarchies of the codes used in the study. They are uploaded to Opasnet Base with the names mentioned after the file name.
- IHME_GBD_2016_CAUSE_HIERARCHY_Y2017M10D02.XLSX → Cause hierarchy (result column: sort_order)
- IHME_GBD_2016_CODEBOOK_Y2017M10D02.CSV → Codebook (result column: lower)
- IHME_GBD_2016_LOCATION_HIERARCHY_Y2017M10D02.XLSX → Location hierarchy (result column: subnational_location)
- IHME_GBD_2016_MEASURE_METRIC_DEFINITIONS_Y2017M10D2.XLSX → Measure metric definitions table on this page
- IHME_GBD_2016_REI_HIERARCHY_Y2017M10D02.XLSX → Rei hierarchy (result column: sort_order)
GBD 2015 study: Causes of death and nonfatal causes mapped to ICD codes [6]
- IHME_GBD_2015_ICD_CAUSE_MAP_CAUSES_OF_DEATH_Y2016M10D07.XLSX → ICD causes of death (result column: ICD9)
- IHME_GBD_2015_ICD_CAUSE_MAP_NONFATAL_Y2016M10D07.XLSX → ICD nonfatal causes (result column: ICD10)
Help for ICD codes: ICD code app, en:ICD-10.

Locations and causes
ICD codes
Combined classification
20.12.2017 Preprocessed data copied to Google sheet [7]

+ Show code - Hide code

# This is code Op_en4685/preprocess on page [[Global Burden of Disase Study]]
library(OpasnetUtils)

bod.disease <- opasnet.csv(
  "0/08/IHME-GBD-2016.zip",
  wiki="opasnet_en",
  unzip="IHME-GBD_2016_DATA-26c7dcb7-1.csv",
  dec=".",
  sep=",",
  header=TRUE
)

bod.country <- opasnet.csv(
  "0/08/IHME-GBD-2016.zip",
  wiki="opasnet_en",
  unzip="IHME-GBD_2016_DATA-d08ee0a4-1.csv",
  dec=".",
  sep=",",
  header=TRUE
)

locations <- opasnet.csv(
  "0/08/IHME-GBD-2016.zip",
  wiki="opasnet_en",
  unzip="IHME_GBD_2016_LOCATION_HIERARCHY_Y2017M10D02.csv",
  dec=".",
  sep=",",
  header=TRUE
)

code <- opasnet.csv(
  "0/08/IHME-GBD-2016.zip",
  wiki="opasnet_en",
  unzip="IHME_GBD_2016_CODEBOOK_Y2017M10D02.CSV",
  dec=".",
  sep=",",
  header=TRUE
)

causes <- opasnet.csv(
  "0/08/IHME-GBD-2016.zip",
  wiki="opasnet_en",
  unzip="IHME_GBD_2016_CAUSE_HIERARCHY_Y2017M10D02.csv",
  dec=".",
  sep=",",
  header=TRUE
)

rei <- opasnet.csv(
  "0/08/IHME-GBD-2016.zip",
  wiki="opasnet_en",
  unzip="IHME_GBD_2016_REI_HIERARCHY_Y2017M10D02.csv",
  dec=".",
  sep=",",
  header=TRUE
)

icdnonfatal <- opasnet.csv(
  "0/08/IHME-GBD-2016.zip",
  wiki="opasnet_en",
  unzip="IHME_GBD_2015_ICD_CAUSE_MAP_NONFATAL_Y2016M10D07.csv",
  dec=".",
  sep=",",
  header=TRUE
)

icddeath <- opasnet.csv(
  "0/08/IHME-GBD-2016.zip",
  wiki="opasnet_en",
  unzip="IHME_GBD_2015_ICD_CAUSE_MAP_CAUSES_OF_DEATH_Y2016M10D07.csv",
  dec=".",
  sep=",",
  header=TRUE
)

#oprint(head(bod.disease))
#oprint(head(bod.country))
#oprint(code)
#oprint(causes)
#oprint(rei) # Something fails does not print
#oprint(icdnonfatal)
#oprint(icddeath)

causes2 <- causes[
  (
    causes$level == 3 | # Take the level 3 diagnoses (168 different diagnoses)
      causes$cause_id %in% c(366, 380, 521)
  ) & # Except only level 2 diagnoses for maternal disorders,
    !causes$parent_id %in% c(366, 380, 521) , # neonatal disorders, and chirrosis and other chronic liver diseases.
  ]
colnames(icdnonfatal)[colnames(icdnonfatal)=="ICD.10"] <- "ICD10"
icddeath$Measure <- "Mortality"
causes.wd <- rbind(
  merge(causes2, icddeath[-3], by.x="cause_name",by.y="Cause", all.x=TRUE),
  merge(causes2, icdnonfatal[-3], by.x="cause_name", by.y="GBD.cause.name", all.x=TRUE)
)

#oprint(causes.wd)

bod.wd <- merge(causes.wd, bod.disease, all.x=TRUE)
bod.wd <- bod.wd[bod.wd$val != 0 , ] # Remove e.g. males with maternal problems

#oprint(bod.wd)

#wp <- re#ad.csv("//cesium/jtue$/_Downloads/query(3).csv")
#causes.wd$cause_name <- tolower(causes.wd$cause_name)
#wp$diseaseLabel <- tolower(wp$diseaseLabel)
#out <- merge(causes.wd, wp, by.x="cause_name", by.y="diseaseLabel", all=TRUE)
#out <- out[order(out$cause_name),]
#View(out[!(is.na(out$cause_id)|is.na(out$ICD10)),])
# Only 10 cause_names shared by Wikipedia and IHME classification

bod <- merge(causes2, bod.disease)
bod <- bod[bod$metric_name != "Percent",]

bod$val <- signif(bod$val, 3)
levels(bod$sex_name)
# [1] "Both"   "Female" "Male"  
levels(bod$sex_name) <- c("", "Female (Q6581072)", "Male (Q6581097)")
levels(bod$location_name)
#[1] "Global"
levels(bod$location_name) <- "Earth (Q2)"

bod$class <- (bod$measure_id-1)*3 + bod$metric_id
oprint(unique(bod[c("class","measure_name","metric_name")]))
#class                           measure_name metric_name
#1     2                                 Deaths     Percent
#2     1                                 Deaths      Number
#4     3                                 Deaths        Rate
#5     6 DALYs (Disability-Adjusted Life Years)        Rate
#8     4 DALYs (Disability-Adjusted Life Years)      Number
#9     5 DALYs (Disability-Adjusted Life Years)     Percent

bod$property <- c( 
  "number of deaths (P1120)", "",
  "incidence (P2844)",
  rep("disease burden (P2854)",3)
)[bod$class]

bod$unit <- c(
  "", "", "cases per 100000 person-years (Q23893296)",
  "disability-adjusted life year (Q55627)","",
  "DALY / 100000 population (Q46123112)"
)[bod$class]

bod$of <- c("", "", "death (Q4)", "", "", "")[bod$class]
bod$ref <- "Global burden of disease study 2016 (Q45750370)"

colnames(bod)
#[1] "cause_id"      "cause_name"    "parent_id"     "level"        
#[5] "cause_outline" "sort_order"    "measure_id"    "measure_name" 
#[9] "location_id"   "location_name" "sex_id"        "sex_name"     
#[13] "age_id"        "age_name"      "metric_id"     "metric_name"  
#[17] "year"          "val"           "upper"         "lower"        
#[21] "class"         "property"      "unit"          "of"           
#[25] "ref"          

bod <- bod[c(2,22,18,23,24,17,10,12,25)]
colnames(bod) <- c(
  "item",
  "property",
  "value",
  "unit",
  "of (P642)",
  "point in time (P585)",
  "location (P276)",
  "sex or gender (P21)",
  "stated in (P248)"
)
bod <- bod[order(bod[[1]],bod[[2]],bod[[4]],bod[[5]],bod[[6]],bod[[8]]),]
oprint(bod, show_all=TRUE)

**Measure metric definitions**
Measure	Number	Percent	Rate	Years
Deaths	Number of deaths in the population	Proportion of deaths for a particular cause relative to deaths from all causes	Deaths per 100,000 population	n/a
Disability adjusted life years (DALYs)	Number of DALYs in the population	Proportion of DALYs for a particular cause relative to DALYs for all causes	DALYs per 100,000 population	n/a
Years of life lost (YLLs)	Number of YLLs in the population	Proportion of YLLs for a particular cause relative to YLDs for all causes	YLLs per 100,000 population	n/a
Years lived with disability (YLDs)	Number of YLDs in the population	Proportion of YLDs for a particular cause relative to YLDs for all causes	YLDs per 100,000 population	n/a
Prevalence	Number of cases in the population	Proportion of cases of a particular cause relative to cases from all causes	Cases per 100,000 population	n/a
Incidence	Number of cases in the population	Proportion of cases of a particular cause relative to cases from all causes	Cases per 100,000 population	n/a
Maternal mortality ratio (MMR)	n/a	n/a	Deaths per 100,000 live births	n/a
Life expectancy	n/a	n/a	n/a	Years lived
Healthy life expectancy (HALE)	n/a	n/a	n/a	Years lived
Summary exposure value (SEV)	n/a	n/a	0 to 100, where 0 is no risk and 100 is the highest level of risk	n/a

2010 Study

Large pieces of data are available at the project website. We are building functionalities to access and reuse the data. Some drafts are written on this page.

An example of R code that fetches mortality data from IHME website and transforms it into a useful format for R-tools. Uncertainties are stored with the mean value for later use with interpret function.

+ Show code - Hide code

# Read data
link <- "http://www.healthmetricsandevaluation.org/sites/default/files/record-attached-files/IHME_GBD_2010_MORTALITY_1970_2010.CSV"
a <- read.csv(link)

# Remove thousand separators.
for(i in 1:ncol(a)) {
	a[[i]] <- gsub(",", "", a[[i]])
}

# Explanations are changed into factors. Observations are characters. 
a[[1]] <- as.factor(a[[1]])
a[[2]] <- as.factor(a[[2]])
a[[3]] <- as.factor(a[[3]])
a[[4]] <- as.factor(a[[4]])
a[[6]] <- paste(a[[5]], a[[6]], sep = " ") # mean with uncertainties
a[[8]] <- paste(a[[7]], a[[8]], sep = " ") # mean with uncertainties
for(i in 1:ncol(a)) print(class(a[, i]))
head(a)
a <- melt(a, id.vars = 1:4, variable_name = "Observation") # Change into long format.

References

Project web page