0% found this document useful (0 votes)
113 views85 pages

GlobalCOD Method 2000 2015

This document describes the methods and data sources used by the World Health Organization to estimate causes of death by country from 2000-2015. It details the process of compiling death registration and other vital statistics data from countries with reliable systems, and modeling causes of death for countries without sufficient data. Specific methods are provided for estimating deaths from diseases and injuries where additional information is available, such as HIV/AIDS, tuberculosis, and cancers. The goal is to produce comprehensive and standardized estimates of mortality by cause for all WHO member states.

Uploaded by

suma riadi
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
113 views85 pages

GlobalCOD Method 2000 2015

This document describes the methods and data sources used by the World Health Organization to estimate causes of death by country from 2000-2015. It details the process of compiling death registration and other vital statistics data from countries with reliable systems, and modeling causes of death for countries without sufficient data. Specific methods are provided for estimating deaths from diseases and injuries where additional information is available, such as HIV/AIDS, tuberculosis, and cancers. The goal is to produce comprehensive and standardized estimates of mortality by cause for all WHO member states.

Uploaded by

suma riadi
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd

 

WHO methods and data sources


for country‐level causes of death
2000‐2015


Department of Information, Evidence and Research
WHO, Geneva

January 2017
 
 
 
 
 
 
 




Global Health Estimates Technical Paper WHO/HIS/IER/GHE/2016.3

Acknowledgments
This Technical Report was written by Colin Mathers, Gretchen Stevens, Wahyu Retno Mahanani, Jessica 
Ho, Doris Ma Fat and Dan Hogan. Estimates of country‐level deaths by cause for years 2000‐2015 were 
primarily  prepared  by  Colin  Mathers,  Gretchen  Stevens,    Jessica  Ho,  Doris  Ma  Fat,  Dan  Hogan  and 
Wahyu  Retno  Mahanani,  of  the  Mortality  and  Health  Analysis  Unit  in  the  WHO  Department  of 
Information, Evidence and Research, in the Health Systems and Innovation Cluster of the World Health 
Organization  (WHO),  Geneva,  drawing  heavily  on  advice  and  inputs  from  other  WHO  Departments, 
collaborating  United  Nations  (UN)  Agencies,  and  WHO  expert  advisory  groups  and  academic 
collaborators.  
Many  of  the  inputs  to  these  estimates  result  from  collaborations  with  Interagency  Groups,  expert 
advisory  groups  and  academic  groups.    These  include  the  Interagency  Group  on  Child  Mortality 
Estimation  (UN‐IGME),  the  UN  Population  Division,  the  Maternal  and  Child  Epidemiology  Estimation 
Group (MCEE), the Maternal Mortality Expert and Interagency Group (MMEIG), the International Agency 
for  Research  on  Cancer,  WHO  QUIVER,  the  Institute  of  Health  Metrics  and  Evaluation  (IHME)  at  the 
University of Washington, and various experts collaborating in the IHME Global Burden of Disease Study. 
While  it  is  not  possible  to  name  all  those  who  provided  advice,  assistance  or  data,  both  inside  and 
outside WHO, we would particularly like to note the assistance and inputs provided by Bob Black, Ties 
Boerma,  Phillipe  Boucher,  Freddie  Bray,  Zoe  Brillantes,  Doris  Chou,  Richard  Cibulskis,  Simon  Cousens, 
Louisa Degenhardt, Brecht  Devleesschauwer, Jeffrey Eaton, Jacques Ferlay, Marta Gacic‐Dobo, Patrick 
Gerland, Arie Havelaar, Stephane Helleringer, Yvan Hutin, Philippe Glaziou, Kacem Iaych, Robert Jakob, 
Prabhat  Jha,  Joy  Lawn,  Li  Liu,  Mary  Mahy,  Bruno  Masquelier,  Shefali  Oza,  Minal  Patel,  Margie  Peden, 
Francois  Pelletier,  Juergen  Rehm,  Florence  Rusciano,  Lale  Say,  Charalampos  Sismanidis,  John  Stover, 
Peter Strebel, Paul Torgerson and Danzhen You. The World Health Organization funded this work. 
Estimates and analysis are available at: 
http://www.who.int/gho/mortality_burden_disease/en/index.html 
For further information about the estimates and methods, or to obtain computer codes, please contact 
[email protected] 
 
In this series
1. CHERG‐WHO methods and data sources for child causes of death 2000‐2015 (Global Health Estimates Technical 
Paper WHO/HIS/HSI/GHE/2016.1) 
2.  WHO  methods  and  data  sources  for  life  tables  1990‐2015  (Global  Health  Estimates  Technical  Paper 
WHO/HIS/IER/GHE/2016.2) 


Table of Contents
 
Acknowledgments ......................................................................................................................................... 2 
Table of Contents .......................................................................................................................................... 3 
1   Introduction 1 
2   Population and all‐cause mortality estimates for years 2000‐2015 ........................................................ 3 
3   Analysis categories ................................................................................................................................... 4 
3.1   Countries .......................................................................................................................................... 4 
3.2   Age groups ....................................................................................................................................... 4 
3.3   Cause of death categories ................................................................................................................ 4 
4   Countries with useable death registration data ...................................................................................... 5 
4.1   Data and estimates .......................................................................................................................... 5 
4.2   Inclusion criteria for countries with high quality death registration data ....................................... 5 
4.3   Mapping to GHE cause lists and redistribution of ill‐defined cause of death ............................... 12 
4.4    Interpolation and extrapolation for missing country‐years........................................................... 20 
4.5   Adjustment of specific causes ........................................................................................................ 21 
5   Causes of death for children under age 5 years .................................................................................... 22 
5.1    Child deaths ................................................................................................................................... 22 
5.2    Child deaths in China ..................................................................................................................... 22 
5.3    Child deaths in India ....................................................................................................................... 23 
6   Causes of death for China 2000‐2015 .................................................................................................... 24 
6.1    Data sources for causes of death ................................................................................................... 24 
6.2    Estimation of deaths by cause for ages 5 and over ....................................................................... 25 
6.3    Comparison of GHE estimates with death registration data and GBD2015 .................................. 26 
7   Causes of death for India 2000‐2015 ..................................................................................................... 27 
7.1    Sample Registration System data .................................................................................................. 27 
7.2    Comparison of GHE estimates with SRS and GBD2015 ................................................................. 27 
8   Methods for specific causes with additional information ..................................................................... 28 
8.1   Tuberculosis ................................................................................................................................... 28 
8.2   HIV/AIDS and sexually transmitted diseases ................................................................................. 28 
8.3   Malaria ........................................................................................................................................... 30 
8.4   Whooping cough ............................................................................................................................ 31 
8.5   Measles .......................................................................................................................................... 31 
8.6   Hepatitis‐attributable deaths ........................................................................................................ 32 
8.7   Schistosomiasis .............................................................................................................................. 34 
8.8   Cycsticercosis, echinococcosis and food‐borne trematodes ......................................................... 34 
8.9  Rabies ............................................................................................................................................. 34 
8.10  Leprosy ........................................................................................................................................... 34 
8.11  Ebola .............................................................................................................................................. 35 
8.12  Maternal causes of death .............................................................................................................. 37 
8.13  Cancers ........................................................................................................................................... 37 
8.14  Alcohol use and drug use disorders ............................................................................................... 37 
8.15  Road injuries .................................................................................................................................. 38 
8.16  Homicide ........................................................................................................................................ 38 
8.17  Conflict and natural disasters ........................................................................................................ 38 
9   Other causes of death for countries without useable data ................................................................... 40 
9.1   Cause of death estimates from the GBD2015 study...................................................................... 40 
9.2   GHECorrect process ....................................................................................................................... 41 
9.3   Other adjustments for specific causes in certain countries .......................................................... 43 
9.4   Comparisons of major cause groups and differences .................................................................... 43 
10  Uncertainty of estimates ....................................................................................................................... 45 
10.1  Guidance on underlying data quality ............................................................................................. 45 
10.2  Uncertainty ranges ......................................................................................................................... 46 
11  Conclusions   .......................................................................................................................................... 52 
11.1   Leading causes of death in 2015 .................................................................................................... 52 
11.2   Reasons for changes in GHE estimates in this revision .................................................................. 55 
11.3   Limitations of GHE estimates ......................................................................................................... 56 
References   .......................................................................................................................................... 58 
Annex Table A  GHE cause categories and ICD‐10 codes ............................................................. 63 
Annex Table B  Groupings of countries, areas and territories used for global and regional tabulations . 69 
B.1    Global ............................................................................................................................................ 69 
B.2   WHO Region .................................................................................................................................. 70 
B.3   World Bank income grouping ........................................................................................................ 71 
B.4   World Bank regions ........................................................................................................................ 72 
Annex Table C GATHER checklist .......................................................................................................... 73 
Annex Table D  Methods used for estimation of mortality levels  and causes of death,  
by country, 2000‐2015 ...................................................................................................... 75 
Annex Table E  First‐level categories for analysis of child causes of death ............................................... 81 
 
 
1 Introduction
Global, regional, and country statistics on population and health indicators are important for assessing 
development and health progress and for guiding resource allocation. The demand is growing for timely 
data to monitor progress on life expectancy and age‐ and cause‐specific mortality rates. This is perhaps 
best  reflected  in  the  indicators  selected  to  monitoring  progress  towards  the  health‐related  targets 
within  the  Sustainable  Development  Goals  (SDGs),  which  will  require  regular  reporting  on  child 
mortality, maternal mortality and mortality due to non‐communicable diseases, suicide, pollution, road 
traffic injuries, homicide, natural disasters and conflict.  
In 2014, WHO published time series estimates of deaths by cause, age and sex for years 2000‐2012 for 
its Member States (1). These have now been updated for years 2000‐2015 drawing on more recent data 
as  summarized  below.  This  technical  paper  documents  the  data  sources  and  methods  used  for 
preparation  of  these  country‐level  Global  Health  Estimates  (GHE)  for  years  2000‐2015.  Annex  Table  A 
lists  the  cause  of  death  categories  and  their  definitions  in  terms  of  the  International  Classification  of 
Diseases, Tenth Revision (ICD‐10) (2). These estimates are available for years 2000, 2005, 2010 and 2015 
for Member States and for selected regional groupings of countries, areas and territories (3), defined in 
Annex Table B, at http://www.who.int/healthinfo/global_health_estimates/en/.  
One of the six core functions of WHO is monitoring of the health situation, trends and determinants in 
the world. Over the years it has cooperated closely with other UN partner agencies like UNICEF, UNAIDS, 
UNFPA  and  the  UN  Population  Division  to  collect  and  compile  global  health  statistics.  There  are  a 
number of established UN multi‐agency expert group mechanisms for  cross cutting topics such as child 
mortality  (the  UN‐IGME  including  UNICEF/WHO/UN  Population  Division/World  Bank),  and  specific 
diseases  such  as  HIV/AIDS  (UNAIDS  Reference  Group),  maternal  mortality  (MMEIG  including 
WHO/UNICEF/UNFPA/World  Bank),  tuberculosis  (WHO  STAG),  malaria  (Malaria  Reference  Group  and 
Roll Back Malaria‐ Malaria Monitoring and Evaluation Reference Group). Additionally, WHO collaborates 
with a network of academics (MCEE) to estimate child causes of death. This collaboration succeeds the 
former Child Health Epidemiology Reference Group (CHERG) of WHO and UNICEF. 
Estimates  of  mortality,  causes  of  death,  DALYs  for  diseases,  injuries  and  risk  factors  were  released  in 
2016  (4‐6)  by  the  Institute  of  Health  Metrics  and  Evaluation  (IHME)  as  part  of  the  Global  Burden  of 
Disease  2015  study  (GBD2015).  WHO  has  drawn  on  the  GBD2015  analyses  for  selected  causes  for 
Member States without comprehensive death registration data as described in Section 9 below.   
These  WHO  Global  Health  Estimates  provide  a  comprehensive  and  comparable  set  of  cause  of  death 
estimates from year 2000 onwards, consistent with and incorporating UN agency, interagency and WHO 
estimates for population, births, all‐cause deaths and specific causes of death, including: 
o most recent vital registration (VR) data for all countries where the VR data quality is assessed as 
useable; 
o updated and additional information on levels and trends for child and adult mortality in many 
countries without good death registration data 
o improvements in methods used for the estimation of causes of child deaths in countries without 
good death registration data.  
o Updated  assessments  of  levels  and  trends  for  specific  causes  of  death  by  WHO  programs  and 
interagency groups. These include: 
 Tuberculosis –WHO 
 HIV – UNAIDS and WHO  

World Health Organization Page 1


 
 Malaria – WHO 
 Vaccine‐preventable child causes – WHO 
 Other major child causes – WHO and CHERG  
 Maternal mortality –MMEIG 
 Cancers – IARC  
 Road traffic accidents – WHO 
 Homicide – WHO  
 Conflict and natural disasters – WHO and the Collaborating Center for Research on the 
Epidemiology of Disasters (CRED) 
o GBD2015  study  estimates  for  other  causes  in  countries  without  useable  VR  data  or  other 
nationally representative sources of information on causes of death. 
Because  these  estimates  draw  on  new  data  and  on  the  result  of  the  GBD2015  study,  and  there  have 
been substantial revisions to methods for many causes, these estimates for the years 2000‐2015 are not 
directly  comparable  with  previous  WHO    estimates  for  2000‐2012  or  earlier  versions.  These  Global 
Health  Estimates  represent  the  best  estimates  of  WHO,  based  on  the  evidence  available  to  it  up  until 
October  2016,  rather  than  the  official  estimates  of  Member  States,  and  have  not  necessarily  been 
endorsed  by  Member  States.  They  have  been  computed  using  standard  categories,  definitions  and 
methods to ensure cross‐national comparability and may not be the same as official national estimates 
produced using alternate, potentially equally rigorous methods. The following sections of this document 
provide explanatory notes on data sources and methods for preparing mortality estimates by cause. 
These estimates have been documented following the Guidelines for Accurate and Transparent Health 
Estimates Reporting (GATHER) (7). The location where GATHER reporting items are reported are given in 
Annex  Table  C.  Figure  1.1  provides  an  overview  of  the  overall  process  of  preparing  the  GHE2015 
estimates  from  the  input  data  sources.  Input  data  and  processes  are  described  in  more  detail  in  the 
following Sections. 
 
   

World Health Organization Page 2


 
Figure  1.1  Overview  of  the  processes  involved  in  the  preparation  of  the  GHE2015  dataset  for 
causes of death in 183 WHO Member States for years 2000‐2015. Refer also to Figure 4.1 for more 
a more detailed summary of the processes involved in the use of death registration data submitted 
to  the  WHO  Mortality  Database  and  to  Figure  9.1  for  a  summary  of  the  data  and  processes 
involved in the preparation of the GHE “prior” estimates dataset. 
 

 


2 Population and all‐cause mortality estimates for years 2000‐2015
WHO life tables have been revised and updated for all Member States for years 1990‐2015, drawing on 
the recently released UN World Population Prospects 2015 revision (8), recent and unpublished analyses 
of all‐cause and HIV mortality for countries with high HIV prevalence, vital registration data (9), and UN‐
IGME estimates of levels and trends for under‐5 mortality (10).  Annex Table D summarizes the methods 
used  for  preparing  life  tables.  Data  sources  are  documented  in  more  detail  in  GHE  Technical  Paper 
2016.2  (11).  The  WHO  life  tables  are  available  in  the  Global  Health  Observatory  at 
http://apps.who.int/gho/data/node.main.LIFECOUNTRY?lang=en 
In  recent  years,  WHO  has  liaised  more  closely  with  the  UN  Population  Division  (on  life  tables  for 
countries, in order to maximize the consistency of UN and WHO life tables, and to minimize differences 
in the use and interpretation of available data on mortality levels. For countries where WHO previously 
predicted  levels  of  adult  mortality  from  estimated  levels  of  child  mortality,  this  update  has  taken  into 

World Health Organization Page 3


 
account additional country‐specific sources of information on levels of adult mortality as reflected in the 
life tables prepared by the UN Population Division for its World Population Prospects (WPP). 
Total deaths by age and sex were estimated for each country by applying the WHO life table death rates 
to  the  estimated  de  facto  resident  populations  prepared  by  the  UN  Population  Division  in  its  2015 
revision (8). They may thus differ slightly from official national estimates for corresponding years.  
 
3 Analysis categories
3.1 Countries
Estimates are made for 183 WHO Member States with populations greater than 90,000 in 2015. The 11 
Member States excluded are: Andorra, Cook Islands, Dominica, Marshall Islands, Monaco, Nauru, Niue, 
Palau, Saint Kitts and Nevis, San Marino, and Tuvalu. This is fewer than the 22 Member States excluded 
for  the  previous  GHE2013  cause  of  death  estimates.  Additionally,  estimates  are  made  for  the  three 
largest  populations  in  non‐Member  State  territories:  Puerto  Rico;  Taiwan,  China;  West  Bank  and  Gaza 
Strip. These are not released at country level, but are included in the relevant regional and global totals.  

3.2 Age groups


The analysis of deaths by cause is carried out for 5‐year age groups from 5‐9, though to the final open‐
ended  age  group  85+.  Deaths  under  age  5  are  estimated  for  the  following  age  groups:  neonatal  (0‐29 
days), postneonatal (1‐11 months), and 1‐4 years. Cause of death estimates are released in tabular form 
for age groups 0‐28 days, 1‐59 months, 5‐14 years, 15‐29, 30‐49, 50‐59, 60‐69, 70+ years.  

3.3 Cause of death categories


The cause of death categories used in the previous WHO cause of death estimates have been expanded 
to  include  a  number  of  additional  causes  and  to  provide  a  more  detailed  breakdown  for  a  several 
causes. The revised GHE2015 cause list is given in Annex Table A, together with corresponding  ICD‐10 
codes.  
New cause categories include: 
Acute hepatitis A 
Acute hepatitis E 
Cysticercosis 
Echinococcosis 
Yellow fever 
Food‐borne trematodosis 
Testicular, kidney, brain, gallbladder, larynx, thyroid cancers and mesothelioma 
Thalassaemias and sickle cell disorders 
Additional digestive disease categories 
Sudden infant death syndrome 
Injuries resulting from unintentional exposure to mechanical forces 
More  detailed  subcategories  have  been  included  for  liver  cancer  and  liver  cirrhosis,  and  for  five 
categories of drug use disorders. The subcategories for liver cancer and liver cirrhosis relate to causes 
including alcohol use and hepatitis infection earlier in life.  

World Health Organization Page 4


 
4 Countries with useable death registration data
4.1 Data and estimates
Cause‐of‐death statistics are reported to WHO on an annual basis by country, year, cause, age and sex. 
These statistics can be accessed in the WHO Mortality Database (9). The number of countries reporting 
data using ICD‐10 has continued to increase. For these estimates, a total of 69 countries had data that 
met  our  inclusion  criteria,  of  which  66 countries  were  reporting  data  coded  to  the  third  or  fourth 
character of ICD‐10 and 61 countries had data for years 2013 or later. Fourteen countries had reported 
data from 2015. 
For countries with a high‐quality vital registration system including information on cause of death, we 
used  the  vital  registration  data  recorded  in  the  WHO  Mortality  Database  to  estimate  cause‐specific 
deaths. We analyzed the data using the following steps: 
1) application of inclusion criteria to select countries with high‐quality vital registration data; 
2) extraction of deaths by cause group, with a short cause list and, if possible, a detailed cause 
list (depending on the cause tabulation used in each country‐year); 
3) redistribution of deaths of unknown sex/age and deaths assigned to ill‐defined (garbage) 
codes and adjustment for incomplete registration of deaths in some countries; 
4) interpolation/extrapolation of number of deaths for missing country‐years; 
5) adjustments to take into account additional information for specific causes of death; and 
6) scaling of total deaths by age and sex to previously estimated WHO all‐cause envelopes for 
years 2000‐2015. 

Figure 4.1 provides an overview of the involved in preparing the complete dataset for GHE causes and 
categories for years 2000 to 2015 for the countries with death registration data reported to the WHO 
Mortality Database and which meet inclusion criteria. Details are provided below.  

4.2 Inclusion criteria for countries with high quality death registration data
We  applied  the  following  inclusion  criteria  to  data  in  the  WHO  mortality  database  received  as  of  end 
October 2016: 
 The data are for a country that is currently a WHO Member State; 
 The data are for a country whose population in 2015 was greater than 90,000; 
 The data are available for 5‐year age groups to ages 85 and over; 
 At least five years of data are available during 2005‐present; 
 The data fulfill quality criteria pertaining to garbage codes and completeness, as described 
below. 
Completeness of death registration data was assessed against estimated total deaths for the population 
as described in the GHE Technical Paper 2016.2 (11). We then calculated the proportion of deaths with 
underlying cause coded to a short list of so‐called “garbage” codes:  
 symptoms, signs and ill‐defined conditions (ICD10 codes R00‐R99), 
 injuries undetermined whether intentional or unintentional (ICD10 Y10‐Y34, Y87.2),  
 ill‐defined cancers (C76, C80, and C97), and  
 ill‐defined cardiovascular diseases ( I47.2, I49.0, I46, I50, I51.4, I51.5, I51.6, I51.9 and I70.9).  

World Health Organization Page 5


 
Figure  4.1  Overview  of  the  processes  involved  in  the  preparation  of  the  GHE2015  dataset  for 
Member States with death registration data meeting inclusion criteria. Refer also to Figure 1.1 for 
further steps involved in the inclusion of this dataset in the final GHE2015 estimates. 

 
 
 
A summary usability score was calculated as follows: 
  (Percent Usable) = Completeness (%) * (1 ‐ Proportion Garbage) 
All  countries  with  a  mean  percent  usable  below  65%  during  the  period  2000  to  latest  available  year 
were excluded (see Table 4.1).   
The quality of cause‐of‐death coding was further investigated in the remaining countries. The proportion 
of deaths assigned to an expanded list of ill‐defined causes (Table 4.2) was calculated for each year in 
the  period  2005‐2014  (or  latest  available),  and  the  mean  proportion  garbage  during  the  period  was 
calculated.  Data from a country were excluded if the average proportion of ill‐defined causes was above 
25%. Based on this analysis, data from Argentina, Bulgaria, Fiji, Greece, Montenegro, Poland, and Syrian 
Arab Republic were excluded (Table 4.1).  
Some data were excluded despite fulfilling our inclusion criteria: from the Philippines, the years 1998‐
1999  and  2002  were  excluded  because  the  trends  in  specific  causes  were  implausible,  and  data  from 
Armenia for the years 2006 and 2008‐2011 were excluded because it was not possible to map the data 
provided to WHO to the shortlist cause list (Table 4.3). Data from South Africa were excluded because of 

World Health Organization Page 6


 
high  levels  of  miscoding  of  HIV  deaths  to  other  causes,  not  captured  by  the  useability  index.  The 
estimation of HIV deaths for South Africa is described in Section 8.2.  
For  countries  which  did  not  meet  the  criteria  for  directly  using  death  registration  data  to  estimate 
causes of death, we have drawn on updated IHME single‐cause analyses from the GBD2015 study (4‐6), 
as described in Section 9. Note that the IHME modelling strategies do make use of the available death 
registration data (9) as well as other sources of information on deaths, covariate regression modelling 
and also draw on patterns of causes of death for similar countries. The country‐specific data and IHME 
analyses can be viewed on their website (12). 
 
 
 
Table 4.1. Characteristicsa of country vital registration data and inclusion/exclusion  
Countryb  Years Available  Average  Range of  Range of garbage  Notes 
usability  completeness  fractionc 
2000‐
latest 
year 
Albania  1998‐2009  60% 76% 85% 18% 28%  Excluded: low 
usability 
Antigua and  1998‐2009, 2012‐ 78% 81% 95% 5% 18%    
Barbuda  2014 
Argentina  1998‐2014  77% 98% 99% 19% 22%  Excluded: high 
proportion 
garbage 
Armenia  1998‐2003, 2012‐ 95% 91% 103% 4% 7%  Excluded: fewer 
2015  than five years' 
data since 2005 
Australia  1998‐2004, 2006‐ 95% 100% 100% 5% 6%    
2014 
Austria  1998‐2014  91% 100% 100% 1% 14%    
Azerbaijan  1998‐2004, 2007  87% 94% 97% 2% 34%  Excluded: fewer 
than five years' 
data since 2005 
Bahamas  1998‐2013  87% 81% 103% 2% 8%    
Bahrain  2001, 2010‐2014  61% 81% 100% 25% 33%  Excluded: low 
usability 
Barbados  2000‐2003, 2005‐ 68% 71% 80% 8% 13%    
2013 
Belarus  1998‐2003, 2007‐ 80% 88% 95% 10% 15%  Summarized 
2011, 2013‐2014  cause list used for 
all years 
Belgium  1998‐2013  87% 100% 100% 12% 15%    
Belize  1998‐2014  76% 74% 90% 4% 13%    
Bolivia  2000‐2003  17% 35% 41% 48% 58%  Excluded: fewer 
(Plurinational  than five years' 
State of)  data since 2005 
Bosnia and  2011, 2014  73% 95% 95% 21% 24%  Excluded: fewer 

World Health Organization Page 7


 
Herzegovina  than five years' 
data since 2005 
Brazil  1998‐2014  83% 93% 100% 10% 21%    
Brunei  1998‐2014  91% 92% 103% 4% 9%  Summarized 
Darussalam  cause list used for 
all years 
Bulgaria  1998‐2013  76% 93% 98% 16% 28%  Excluded: high 
proportion 
garbage 
Cabo Verde  2011  73% 95% 95% 23% 23%  Excluded: fewer 
than five years' 
data since 2005 
Canada  1998‐2012  94% 100% 100% 6% 8%    
Chile  1998‐2014  94% 99% 100% 5% 11%    
Colombia  1998‐2013  84% 80% 91% 5% 8%    
Costa Rica  1998‐2014  87% 91% 98% 4% 7%    
Croatia  1998‐2015  89% 100% 100% 6% 17%    
Cuba  1998‐2014  92% 98% 100% 1% 9%    
Cyprusd  2004‐2013  58% 66% 76% 10% 24%    

Czechia  1998‐2015  89% 100% 100% 7% 15%    

Denmark  1998‐2014  87% 100% 100% 12% 14%    


Dominican  1998‐2012  45% 50% 61% 8% 21%  Excluded: low 
Republic  usability 
Ecuador  1998‐2014  69% 79% 93% 14% 23%    
Egypt  2000‐2011  56% 89% 94% 32% 41%  Excluded: low 
usability 
El Salvador  1998‐2013  63% 75% 84% 18% 26%  Excluded: low 
usability 
Estonia  1998‐2014  94% 100% 100% 5% 8%    
Fiji  2001‐2009, 2011‐ 81% 103% 103% 9% 37%  Excluded: high 
2012  proportion 
garbage 
Finland  1998‐2014  97% 100% 100% 2% 3%    
France  1998‐2013  85% 100% 100% 14% 16%    
Georgia  1998‐2001, 2004‐ 61% 81% 100% 7% 69%  Excluded: low 
2007, 2009‐2014  usability 
Germany  1998‐2014  88% 100% 100% 11% 14%    
Greece  1998‐2013  74% 97% 100% 24% 27%  Excluded: high 
proportion 
garbage 
Grenada  2001‐2015  88% 87% 100% 5% 15%    
Guatemala  1998‐2014  79% 90% 96% 9% 22%    
Guyana  1998‐1999, 2001‐ 86% 76% 93% 6% 22%    
2012 
Haiti  1999, 2001‐2004  6% 3% 15% 32% 52%  Excluded: fewer 
than five years' 

World Health Organization Page 8


 
data since 2005
Honduras  2008‐2013  13% 13% 15% 3% 7%  Excluded: low 
usability 
Hungary  1998‐2014  95% 100% 100% 4% 7%    
Iceland  1998‐2015  94% 100% 100% 5% 10%    
Iraq  2008  54% 75% 75% 28% 28%  Excluded: fewer 
than five years' 
data since 2005 
Ireland  1998‐2013  95% 100% 100% 4% 8%  Summarized 
cause list used for 
some years 
Israel  1998‐2014  91% 100% 100% 8% 14%    
Italy  1998‐2003, 2006‐ 91% 100% 100% 8% 12%    
2012 
Jamaica  2000‐2006, 2009‐ 73% 73% 92% 5% 25%    
2011 
Japan  1998‐2014  89% 100% 100% 9% 15%    
Jordan  2008‐2011  61% 59% 86% 9% 10%  Excluded: fewer 
than five years' 
data since 2005 
Kazakhstan  1998‐2015  81% 84% 90% 3% 22%  Summarized 
cause list used for 
all years 
Kiribati  1998‐2001  51% 68% 79% 25% 35%  Excluded: fewer
than five years' 
data since 2005 
Kuwait  1998‐2014  65% 63% 86% 8% 16%  Excluded: low 
usability 
Kyrgyzstan  1998‐2015  90% 92% 97% 3% 8%    
Latvia  1998‐2014  90% 95% 98% 5% 11%    
Lithuania  1998‐2015  92% 91% 99% 2% 6%    
Luxembourg  1998‐2014  86% 100% 100% 12% 16%    
Malaysia  2000‐2008  40% 46% 58% 21% 24%  Excluded: fewer 
than five years' 
data since 2005 
Maldives  2000‐2005, 2007‐ 58% 84% 102% 13% 77%  Excluded: low 
2008, 2010‐2011  usability 
Malta  1998‐2014  92% 100% 100% 5% 12%    
Mauritius  1998‐2014  88% 96% 97% 7% 15%    
Mexico  1998‐2014  95% 100% 100% 4% 6%    
Montenegro  2000‐2009  67% 89% 94% 23% 29%  Excluded: high 
proportion 
garbage 
Morocco  2008‐2012  12% 21% 24% 46% 51%  Excluded: low 
usability 
Netherlands  1998‐2015  86% 100% 100% 13% 15%    
New Zealand  1998‐2012  97% 100% 100% 3% 4%    
Nicaragua  1998‐2013  60% 49% 72% 4% 11%  Excluded: low 
usability 

World Health Organization Page 9


 
Norway  1998‐2014  88% 100% 100% 11% 13%    
Oman  2009‐2010  49% 81% 84% 34% 46%  Excluded: fewer 
than five years' 
data since 2005 
Panama  1998‐2014  80% 86% 93% 8% 14%    
Paraguay  1998‐2014  61% 66% 83% 14% 27%  Excluded: low 
usability 
Peru  1998‐2014  57% 61% 68% 5% 24%  Excluded: low 
usability 
Philippines  2000‐2001, 2003,  78% 85% 90% 9% 10%    
2006‐2011 
Poland  1999‐2014  72% 98% 100% 25% 31%  Excluded: high 
proportion 
garbage 
Portugal  1998‐2003, 2007‐ 82% 100% 100% 14% 22%  Summarized 
2014  cause list used for 
some years 
Qatar  2001, 2004‐2012  60% 65% 97% 22% 35%  Excluded: low 
usability 
Republic of  1998‐2013  85% 97% 100% 13% 21%    
Korea 
Republic of  1998‐2015  83% 80% 90% 2% 7%    
Moldova 
Romania  1998‐2015  92% 100% 100% 0% 8%    
Russian  1998‐2011  89% 91% 96% 4% 6%  Summarized 
Federation  cause list used for 
all years 
Saint Lucia  1998‐2006, 2008‐ 75% 76% 94% 6% 27%    
2014 
Saint Vincent  1998‐2015  93% 92% 104% 2% 10%    
and the 
Grenadines 
Saudi Arabia  2009, 2012  21% 39% 39% 46% 48%  Excluded: fewer 
than five years' 
data since 2005 
Serbia  1998‐2014  80% 93% 96% 12% 18%    
Singapore  1998‐2015  71% 68% 82% 1% 4%    
Slovakia  1998‐2010, 2012‐ 94% 100% 100% 4% 11%    
2014 
Slovenia  1998‐2013, 2015  89% 100% 100% 9% 12%    
South Africa  1998‐2014  70% 81% 97% 19% 32%  Special methods 
used 
Spain  1998‐2014  90% 100% 100% 8% 12%    
Sri Lanka  1998‐2003, 2006  72% 79% 102% 23% 32%  Excluded: fewer 
than five years' 
data since 2005 
Suriname  1998‐2014  64% 66% 79% 12% 22%  Excluded: low 
usability 
Sweden  1998‐2015  89% 100% 100% 10% 12%    
Switzerland  1998‐2013  89% 100% 100% 10% 13%    

World Health Organization Page 10


 
Syrian Arab  1998‐2010  72% 79% 103% 10% 35%  Excluded: high 
Republic  proportion 
garbage 
Tajikistan  1998‐2005  78% 78% 83% 4% 9%  Excluded: fewer 
than five years' 
data since 2005 
Thailand  1998‐2000, 2002‐ 52% 79% 95% 31% 54%  Excluded: low 
2014  usability 
The former  1998‐2013  87% 96% 103% 9% 16%  Summarized 
Yugoslav  cause list used for 
Republic of  some years 
Macedonia 
Trinidad and  1998‐2010  84% 81% 100% 2% 5%    
Tobago 
Tunisia  2009, 2013  22% 28% 30% 18% 27%  Excluded: fewer 
than five years' 
data since 2005 
Turkey  1999‐2002, 2004‐ 58% 47% 91% 8% 15%  Excluded: low 
2013  usability 
Turkmenistan  1998, 2012‐2013  67% 77% 94% 3% 13%  Excluded: fewer 
than five years' 
data since 2005 
Ukrained  1998‐2012, 2014  90% 89% 97% 2% 6%  Summarized 
cause list used for 
all years 
United Arab  2005‐2010  56% 59% 82% 18% 26%  Excluded: low 
Emirates  usability 
United  1998‐2014  93% 100% 100% 6% 8%    
Kingdom 
United States of  1998‐2014  93% 100% 100% 7% 10%    
America 
Uruguay  1998‐2010, 2012‐ 83% 100% 100% 16% 18%    
2014 
Uzbekistan  1998‐2005, 2009‐ 85% 81% 99% 2% 6%  Summarized 
2014  cause list used for 
some years 
Venezuela  1998‐2013  82% 86% 90% 7% 9%    
(Bolivarian 
Republic of) 
a) Characteristics on data sources that are common to all sources are not listed in this table.  Specifically, all 
data  sources  cover  the  national  area  unless  otherwise  noted,  are  death  registration  data  based  on 
medical certification of death, and cover all ages and both sexes.  
b) Only data fulfilling the first three inclusion criteria listed above, e.g. member state, minimum population 
and detailed age grouping, are included in this table.  
c) ICD‐10 codes included in the “garbage” category are given in the text above. Additional codes in Table 4.2 
are not considered in this column. 
d) Data are for areas under government control. 
 

World Health Organization Page 11


 
Table 4.2. Expanded list of garbage codes 

ICD‐10 code(s)  Description 
A40‐A41  Streptococcal  and other septicaemia 
C76, C80, C97     Ill‐defined cancer sites 
D65  Disseminated intravascular coagulation [defibrination syndrome] 
E86  Volume depletion 
I10  Essential (primary) hypertension 
I269  Pulmonary embolism without mention of acute cor pulmonale 
I46  Cardiac arrest 
I472  Ventricular tachycardia 
I490  Ventricular fibrillation and flutter 
I50  Heart failure 
I514 Myocarditis, unspecified 
I515 Myocardial degeneration 
I516 Cardiovascular disease, unspecified 
I519 Heart disease, unspecified 
I709 Generalized and unspecified atherosclerosis 
I99 Other and unspecified disorders of circulatory system 
J81 Pulmonary oedema 
J96  Respiratory failure, not elsewhere classified 
K72 Hepatic failure, not elsewhere classified 
N17  Acute renal failure 
N18 Chronic renal failure 
N19  Unspecified renal failure 
P285 Respiratory failure of newborn 
Y10‐Y34, Y872   External cause of death not specified as accidentally or purposely inflicted 
 

4.3 Mapping to GHE cause lists and redistribution of ill‐defined cause of death
Included  vital  registration  data  were  coded  according  to  ICD9,  ICD10,  or  one  of  several  abbreviated 
cause lists derived  from ICD9  or ICD10.   Total  deaths by  cause, age  and sex  were mapped  to the  GHE 
cause  list  (Annex  Table  A).  We  used  the  complete  cause  list  in  Annex  Table  A  if  the  data  were  coded 
using 3‐ or 4‐digit ICD‐10 codes or 4‐digit ICD‐9 codes.  For all included data, we extracted the number of 
deaths by cause, age and sex, using the broad cause categories listed in Table 4.3 (hereafter “shortlist”).  
In some cases, counts of deaths were not available for specific causes of death. Specifically, chlamydia 
deaths were not available in the  4‐digit ICD‐9 codes.  The  mean fraction of other sexually  transmitted 
disease deaths caused by chlamydia was calculated for each country‐sex group and applied to all years 
of data for that country.  If there were no deaths coded to other sexually transmitted diseases in a given 
country, the mean fraction for all other countries was used. Several causes of death are not available in 
death registration data  coded using ICD10  at the 3‐digit  level: hepatitis C  (acute infections),  lymphatic 
filariasis, Japanese encephalitis, panic disorder, age‐related vision disorders, congenital abdominal wall 
defect, and congenital oesophageal atresia. Deaths for all of these causes were assumed to be zero in 
the countries with data coded to ICD10 at the 3‐digit level. 

World Health Organization Page 12


 
Deaths of unknown sex were redistributed pro‐rata within cause‐age groups of known sexes, and then 
deaths  of  unknown  age  were  redistributed  pro‐rata  within  cause‐sex  groups  of  known  ages.  We 
redistributed  deaths  coded  to  symptoms,  signs  and  ill‐defined  conditions  (ICD10  codes  R00‐R94,R96‐
R99) pro‐rata to all non‐injury causes of death, and injuries with undetermined intent (ICD10 codes Y10‐
Y34) pro‐rata to all injury causes of death, following previously published methods (13).  Cancers with 
unspecified  site  (ICD10  codes  C76,  C80,  C97)  were  redistributed  pro‐rata  to  all  sites  excluding  liver, 
pancreas, ovary, and lung.  Additionally, we redistributed cancer of uterus, part unspecified (C55) pro‐
rata to cervix uteri (C53) and corpus uteri (C54). 

Previously  published  analyses  of  heart  failure  (14,  15)  have  proposed  that  these  deaths  be  reassigned 
mainly  to  to  ischemic  heart  disease  (IHD;  cause  1100),  chronic  obstructive  pulmonary  disease  (COPD; 
cause  1180)  in  older  adults,  and  to  IHD,  COPD,  cardiomyopathy,  myocarditis,  and  endocarditis  (cause 
1150)  and  congenital  heart  anomalies  (cause  1440)  in  children,  adolescents  and  young  adults 
(destination  causes  for  ill‐defined  deaths  may  be  called  target  causes).    Following  these  analyses,  we 
redistributed  heart  failure  and  other  ill‐defined  cardiovascular  causes  of  death  to  IHD  and  COPD  in 
adults  over  age  50  and  to  the  four  target  causes—IHD,  COPD,  cardiomyopathy,  myocarditis, 
endocarditis, and congenital heart anomalies in people under age 50. As these conditions have strong 
age and sex patterns, redistribution fractions were calculated by age and sex.  We combined available 
data  from  three  epidemiologically  relevant  regions,  the  traditional  high‐income  countries,  Eastern 
Europe  and  Central  Asia,  and  other  countries  with  usable  death  registration  data,  and  calculated 
fractions  for  each  target  disease  based  on  their  relative  frequency  in  the  data.    The  redistribution 
fractions are shown in Tables 4.4‐6. 
The  ICD‐10  code  ranges  mapped  to  hypertensive  heart  disease  (HHD)  include  codes  for  essential 
hypertension  (I10),  secondary  hypertension  (I15)  and  hypertensive  renal  disease  (I12).  Most  deaths 
coded  to  essential  hypertension  are  likely  to  be  due  to  ischaemic  heart  disease,  and  additionally  it  is 
likely that a proportion of deaths coded to HHD  are actually due to ischaemic heart disease in  people 
who also had essential hypertension.  
Based on a regression analysis of the logit of the proportion of deaths in the  HHD category that were 
coded to essential hypertension against the crude HHD death rate, the predicted fraction of HHD deaths 
to be redistributed to IHD was estimated. It was set to 30% for country‐years with HHD death rate less 
than 20 per 100,000.  For certain outlier countries, it was set to country‐specific values derived directly 
from the VR data: 50% (Brazil), 40% (France) and 37% (Argentina).  Based on a similar analysis, 10% of 
HHD deaths were redistributed to “other chronic kidney disease”, with specific higher values for Japan 
(18%), Mexico (30%) and the USA (15%). 
In  a  number  of  countries,  the  deaths  coded  to  the  GHE  category  “other  infectious  diseases”  result  in 
unusually high death rates for this category. The GHE category includes a number of within‐infectious‐
disease garbage codes: A49  Bacterial infection, unspecified; A89  Unspecified viral infection of the CNS; 
B34  Viral infection of unspecified site; B94  Sequelae of other and unspecified infectious disease; and 
B99    Other  and  unspecified  infectious  diseases.  However,  the  numbers  of  deaths  coded  to  these 
categories  are  insufficient  to  explain  the  outlier  levels.    Based  on  a  regression  the  death  rate  for  this 
category  against  the  death  rate  for  lower  respiratory  infections,  fractions  of  the  “other  infectious 
disease” deaths were shifted to lower respiratory infections for all except shortlist countries (see Table 
4.7).  The  average  fraction  was  13%  with  20th  percentile  4%  and  80th  percentile  29%.    Countries  with 
average  fractions  of  30%  or  more  included  Antigua  and  Barbuda,  Belgium,  Barbados,  Grenada,  Israel, 
Saint Lucia, Luxembourg, Norway, Sweden, Switzerland and St. Vincent and the Grenadines.  
   

World Health Organization Page 13


 
Table 4.3. Short cause list used for vital registration data coded using ICD‐9 or ICD‐10 abbreviated cause 
lists 
GHE 
code  Shortlist cause category 
10  I.  Communicable, maternal, perinatal and nutritional conditions 
20  A. Infectious and parasitic diseases
30    A1. Tuberculosis 
100    A3. HIV/AIDS 
220           A9a. Malaria 
380  B. Respiratory infections 
390    B1. Lower respiratory infections
420  C. Maternal conditions 
490  D. Neonatal conditions 
540  E. Nutritional deficiencies 
600  II. Noncommunicable diseases 
610  A. Malignant neoplasms 
620    A1. Mouth and oropharynx cancers
630    A2. Oesophagus cancer 
640    A3. Stomach cancer 
650    A4. Colon and rectum cancers
660    A5. Liver cancer 
680    A7. Trachea, bronchus and lung cancers
700    A9. Breast cancer 
710    A10. Cervix uteri cancer 
740    A13.  Prostate cancer 
800  C. Diabetes mellitus 
820/  E/F. Mental and neurological disorders
940 
1100  H. Cardiovascular diseases 
1130     H3. Ischaemic heart disease 
1140     H4. Stroke 
1170  I. Respiratory diseases 
1180     I1. Chronic obstructive pulmonary disease
1190     I2. Asthma 
1200     I3. Other respiratory diseases 
1210  J. Digestive disorders 
1230     J2. Liver cirrhosis 
1260  K. Genitourinary diseases 
1400  N. Congenital anomalies 
1510  III. Injuries 
1520  A. Unintentional injuries 
1530    A1. Road injury 
1600  B. Intentional injuries 
1610    B1. Self‐harm
1620    B2. Interpersonal violence 
1630    B3. Collective violence and legal intervention
 

World Health Organization Page 14


 
Table  4.4.  Redistribution  fractions  for  ill‐defined  cardiovascular  causes  of  death  (ICD10  4‐digit  codes 
I472, I490, I46, I50, I514, I515, I516, I519, and I709) for the traditionally high‐income countriesa 
GHE target cause 

  1100  1150  1180  1440  1100  1150  1180  1440 


Age  Redistribution fractions for males  Redistribution fractions for females 

0  1%  6%  1% 93% 1% 7%  1%  92%


1‐4  2%  19%  4% 75% 2% 22%  3%  73%
5‐9  4%  26%  4% 66% 4% 32%  4%  60%
10‐14  6%  35%  3% 55% 5% 34%  4%  57%
15‐19  14%  42%  3% 41% 11% 37%  5%  48%
20‐24  29%  42%  3% 26% 20% 41%  4%  35%
25‐29  46%  35%  3% 16% 33% 37%  4%  25%
30‐39  63%  27%  2% 8% 48% 31%  5%  16%
35‐39  74%  19%  3% 4% 61% 24%  6%  9%
40‐44  81%  14%  4% 2% 69% 17%  10%  5%
45‐49  83%  10%  5% 1% 70% 13%  14%  3%
50‐54  91%    9%   78%   22%   
55‐59  88%    12%   73%   27%   
60‐64  84%    16%   71%   29%   
65‐69  80%    20%   70%   30%   
70‐74  76%    24%   71%   29%   
75‐79  74%    26%   75%   25%   
80‐84  74%    26%   79%   21%   
85+  77%    23%   86%   14%   
a) Andorra,  Australia,  Austria,  Belgium,  Canada,  Cyprus,  Denmark,  Finland,  France,  Germany,  Greece, 
Iceland,  Ireland,  Israel,  Italy,  Japan,  Luxembourg,  Malta,  Monaco,  Netherlands,  New  Zealand,  Norway, 
Portugal, Republic of Korea, San Marino, Singapore, Spain, Sweden, Switzerland, United Kingdom, United 
States of America 

   

World Health Organization Page 15


 
Table  4.5.  Redistribution  fractions  for  ill‐defined  cardiovascular  causes  of  death  (ICD10  4‐digit  codes 
I472, I490, I46, I50, I514, I515, I516, I519, and I709) for eastern European and central Asian countriesa 
GHE target cause 

  1100  1150  1180  1440  1100  1150  1180  1440 


Age  Redistribution fractions for males  Redistribution fractions for females 

0  0%  2%  0%  97%  0%  3%  0%  97% 


1‐4  2%  9%  3%  86%  2%  12%  2%  84% 
5‐9  4%  20%  5%  71%  2%  23%  4%  71% 
10‐14  5%  29%  8%  57%  5%  27%  11%  57% 
15‐19  21%  34%  9%  37%  18%  26%  10%  46% 
20‐24  50%  27%  10%  14%  47%  19%  14%  20% 
25‐29  59%  26%  8%  7%  58%  21%  10%  11% 
30‐39  66%  25%  7%  3%  59%  23%  11%  7% 
35‐39  72%  21%  6%  1%  66%  21%  9%  4% 
40‐44  76%  17%  6%  1%  73%  17%  8%  2% 
45‐49  80%  13%  7%  0%  76%  14%  9%  1% 
50‐54  92%    8%    89%    11%   
55‐59  90%    10%    89%    11%   
60‐64  88%    12%    90%    10%   
65‐69  87%    13%    91%    9%   
70‐74  86%    14%    92%    8%   
75‐79  86%    14%    92%    8%   
80‐84  87%    13%    93%    7%   
85+  90%    10%    94%    6%   
a) Albania,  Armenia,  Azerbaijan,  Belarus,  Bosnia  and  Herzegovina,  Bulgaria,  Croatia,  Czechia,  Estonia, 
Georgia, Hungary, Kazakhstan, Kyrgyzstan, Latvia, Lithuania, Mongolia, Montenegro, Poland, Republic of 
Moldova,  Romania,  Russian  Federation,  Serbia,  Slovakia,  Slovenia,  Tajikistan,  The  former  Yugoslav 
Republic of Macedonia, Turkmenistan, Ukraine, Uzbekistan 
 

   

World Health Organization Page 16


 
Table  4.6.  Redistribution  fractions  for  ill‐defined  cardiovascular  causes  of  death  (ICD10  4‐digit  codes 
I472, I490, I46, I50, I514, I515, I516, I519, and I709) for all other countries 
 
GHE target cause 

  1100  1150  1180  1440  1100  1150  1180  1440 


Age  Redistribution fractions for males  Redistribution fractions for females 

0  0%  3%  1% 95% 0% 3%  1%  95%


1‐4  1%  9%  5% 84% 1% 9%  4%  85%
5‐9  4%  15%  5% 76% 3% 15%  4%  77%
10‐14  9%  23%  5% 63% 7% 22%  5%  66%
15‐19  38%  23%  6% 33% 30% 22%  6%  42%
20‐24  59%  20%  5% 15% 44% 23%  6%  26%
25‐29  69%  18%  5% 8% 54% 22%  7%  17%
30‐39  75%  16%  5% 4% 65% 18%  7%  9%
35‐39  79%  14%  5% 2% 72% 15%  8%  5%
40‐44  82%  11%  6% 1% 76% 12%  9%  2%
45‐49  83%  9%  7% 1% 78% 10%  11%  1%
50‐54  89%    11%   85%   15%   
55‐59  86%    14%   82%   18%   
60‐64  81%    19%   80%   20%   
65‐69  76%    24%   77%   23%   
70‐74  71%    29%   75%   25%   
75‐79  68%    32%   74%   26%   
80‐84  66%    34%   74%   26%   
85+  66%    34%   75%   25%   
 
 
GHE categories 950 “Alzheimer disease and other dementias” and 1010 “Other neurological conditions” 
contain  84%  of  the  deaths  coded  to  neurological  causes  in  the  death  registration  data  for  2000‐2015. 
“Other  neurological  conditions”  accounted  for  15%  on  average,  but  in  some  countries  accounted  for 
much  higher  proportions  of  deaths,  eg.  Uzbekhistan  63%,  Guatemala  59%,  Singapore  52%,  Colombia 
52%,  Philippines  46%,  Mexico  44%,  Brazil  30%.  Based  on  a  regression  of  the  log  of  the  “Other 
neurological  conditions”  death  rate  against  the  log  of  the  death  rate  for  dementias,  excess  “other 
neurological” deaths above the predicted rate were shifted to the dementia category (see Table 4.7).  

World Health Organization Page 17


 
Table 4.7. Fractions of selected “other” categories redistributed as described above. Hypertensive heart 
disease (HHD) fractions refer to redistribution to ischaemic heart disease (IHD) and “other chronic kidney 
disease (CKD) 
ISO3  Member State  Other     Other neurological  Other              HHD  
infectious     male          female  respiratory         IHD             CKD 
ATG  Antigua and Barbuda  0.378  0.551  0.078  0.374  0.373  0.10 
AUS  Australia  0.150  0.198  0.223  0.261  0.300  0.10 
AUT  Austria  0.176  0.226  0.175  0.000  0.334  0.10 
BEL  Belgium  0.386  0.188  0.270  0.253  0.300  0.10 
BHS  Bahamas  0.245  0.000  0.000  0.365  0.382  0.10 
BLR  Belarus  0.050  0.000  0.000  0.000  0.300  0.10 
BLZ  Belize  0.079  0.549  0.199  0.309  0.334  0.10 
BRA  Brazil  0.132  0.167  0.000  0.287  0.500  0.10 
BRB  Barbados  0.500  0.181  0.150  0.372  0.382  0.10 
BRN  Brunei Darussalam  0.037  0.000  0.000  0.000  0.300  0.10 
CAN  Canada  0.078  0.131  0.176  0.232  0.300  0.10 
CHE  Switzerland  0.388  0.169  0.309  0.184  0.327  0.10 
CHL  Chile  0.027  0.000  0.000  0.305  0.306  0.10 
COL  Colombia  0.044  0.565  0.537  0.134  0.300  0.10 
CRI  Costa Rica  0.156  0.129  0.000  0.206  0.300  0.10 
CUB  Cuba  0.042  0.000  0.000  0.000  0.305  0.10 
CYP  Cyprus  0.131  0.149  0.087  0.353  0.309  0.10 
CZE  Czechia  0.052  0.111  0.000  0.104  0.303  0.10 
DEU  Germany  0.265  0.094  0.076  0.200  0.337  0.10 
DNK  Denmark  0.218  0.000  0.000  0.000  0.300  0.10 
ECU  Ecuador  0.045  0.482  0.394  0.319  0.321  0.10 
ESP  Spain  0.125  0.000  0.000  0.334  0.302  0.10 
EST  Estonia  0.056  0.289  0.023  0.179  0.486  0.10 
FIN  Finland  0.087  0.000  0.000  0.167  0.302  0.10 
FRA  France  0.272  0.276  0.283  0.342  0.400  0.10 
GBR  United Kingdom  0.067  0.000  0.000  0.265  0.300  0.10 
GRD  Grenada  0.397  0.468  0.347  0.347  0.352  0.10 
GTM  Guatemala  0.014  0.691  0.687  0.330  0.300  0.10 
GUY  Guyana  0.290  0.461  0.450  0.330  0.350  0.10 
HRV  Croatia  0.002  0.424  0.542  0.000  0.320  0.10 
HUN  Hungary  0.131  0.000  0.000  0.000  0.387  0.10 
IRL  Ireland  0.109  0.042  0.055  0.244  0.300  0.10 
ISL  Iceland  0.138  0.000  0.106  0.000  0.300  0.10 
ISR  Israel  0.369  0.296  0.313  0.316  0.300  0.10 
ITA  Italy  0.128  0.136  0.246  0.235  0.352  0.10 
JAM  Jamaica  0.236  0.000  0.000  0.261  0.350  0.10 
JPN  Japan  0.200  0.345  0.000  0.367  0.300  0.18 
KAZ  Kazakhstan  0.000  0.000  0.000  0.000  0.301  0.10 
KGZ  Kyrgyzstan  0.067  0.655  0.675  0.000  0.300  0.10 

World Health Organization Page 18


 
KOR  Republic of Korea  0.012  0.000  0.000  0.286  0.300  0.10 
LCA  Saint Lucia  0.402  0.474  0.406  0.343  0.354  0.10 
LTU  Lithuania  0.039  0.277  0.033  0.287  0.301  0.10 
LUX  Luxembourg  0.326  0.000  0.075  0.000  0.300  0.10 
LVA  Latvia  0.151  0.288  0.000  0.239  0.336  0.10 
MDA  Republic of Moldova  0.025  0.265  0.317  0.155  0.313  0.10 
MEX  Mexico  0.059  0.453  0.335  0.000  0.300  0.30 
MKD  The former Yugoslav  0.000  0.389  0.296  0.166  0.337  0.10 
Republic of Macedonia 
MLT  Malta  0.043  0.000  0.000  0.298  0.300  0.10 
MUS  Mauritius  0.137  0.385  0.269  0.278  0.328  0.10 
NLD  Netherlands  0.233  0.042  0.095  0.370  0.300  0.10 
NOR  Norway  0.305  0.051  0.081  0.083  0.300  0.10 
NZL  New Zealand  0.054  0.000  0.000  0.022  0.300  0.10 
PAN  Panama  0.009  0.476  0.332  0.009  0.300  0.10 
PHL  Philippines  0.196  0.475  0.444  0.295  0.322  0.10 
PRT  Portugal  0.263  0.000  0.000  0.338  0.302  0.10 
ROU  Romania  0.027  0.000  0.000  0.112  0.516  0.10 
RUS  Russian Federation  0.000  0.000  0.000  0.000  0.300  0.10 
SGP  Singapore  0.070  0.562  0.525  0.190  0.300  0.10 
SRB  Serbia  0.053  0.366  0.343  0.231  0.347  0.10 
SVK  Slovakia  0.037  0.000  0.000  0.211  0.343  0.10 
SVN  Slovenia  0.025  0.346  0.126  0.139  0.316  0.10 
SWE  Sweden  0.333  0.000  0.000  0.150  0.302  0.10 
TTO  Trinidad and Tobago  0.219  0.265  0.000  0.330  0.336  0.10 
UKR  Ukraine  0.000  0.000  0.000  0.000  0.300  0.10 
URY  Uruguay  0.221  0.461  0.618  0.305  0.301  0.10 
USA  United States of America  0.295  0.161  0.213  0.208  0.302  0.15 
UZB  Uzbekistan  0.006  0.696  0.714  0.234  0.382  0.10 
VCT  Saint Vincent & the  0.428  0.591  0.053  0.354  0.374  0.10 
Grenadines 
VEN  Venezuela   0.068  0.280  0.035  0.130  0.300  0.10 

Similar issues occurred for chronic respiratory disease categories, with high proportions of deaths coded 
to  “other  respiratory  diseases”  in  some  countries.  Denoting  the  fraction  chronic  respiratory  disease 
deaths  in  the  “other  respiratory  diseases”  category  as  rfrac,  for  countries  where  rfrac  exceeded  the 
initial  average  value  of  0.15,    it  was  rescaled  to  fall  in  the  range  0.15  to  0.5  (one  standard  deviation 
above  the  mean).  The  excess  deaths  in  the  “other”  category  were  shifted  pro‐rata  by  age  and  sex  to 
COPD and asthma cause categories (see Table 4.7).  

World Health Organization Page 19


 
4.4 Interpolation and extrapolation for missing country‐years
For many countries, data were missing for some years.  In order to create a continuous time‐series of 
data  from  2000  to  2015,  we  interpolated  mortality  rates  for  each  country  and  cause,  and  then 
extrapolated  up  to  six  years  of  data  at  the  beginning  and  end  of  the  data  series.  Interpolation  and 
extrapolation was carried out separately for the detailed cause list and the short cause list. All shortlist 
interpolations and extrapolations were carried out using all available data meeting the inclusion criteria. 
A description of the methods follows. 
For each country‐age‐sex‐cause group of the detailed cause list: 
1) We  interpolated  by  calculating  the  mean  death  rate  of  all  available  data  in  a  seven‐year 
window (three years on either side, no earlier than 1998).
2) We extrapolated up to six years from the first/last year of data by applying the mean death 
rate from the first three or last three years of data to the missing data‐years.
For each country‐age‐sex‐cause group of the shortlist cause list: 
1) We  interpolated  by  fitting  a  logistic  regression  for  each  missing  country‐age‐sex‐cause 
group,  using  death  rates  six  years  prior  (but  no  earlier  than  1998)  and  six  years  after  the 
missing data year as the dependent variable and year as the independent variable.  In some 
cases, few deaths were recorded for a specific country‐age‐sex‐cause group and the logistic 
regression did not converge.  In that case, the death rate was estimated as the average rate 
in the three years prior and three years following the missing data year (as was done for the 
detailed cause list).  
2) Extrapolation method depended on mean number of deaths in the first/last three years of 
data:
a. If there were an average of more than 250 deaths, a logistic regression was fitted to 
the  first  or  the  final  six  years  of  data  (including  interpolated  estimates)  for  each 
country‐sex‐cause. 
b. If there were an average 250 or fewer deaths, we extrapolated up to six years from 
the  first/last  year  of  data  by  applying  the  mean  death  rate  from  the  first  three  or 
last  three  years  of  data  to  the  missing  data‐years  (as  was  done  for  the  detailed 
cause list).
 
Because more shortlist data were available than detailed list data, and shortlist data were interpolated 
and extrapolated using regression methods that reflect trends in death rates, deaths by cause according 
to the detailed cause list were adjusted to sum to the totals in the filled‐in shortlist dataset. This implied 
no change when the detailed cause list data were available (most country‐years).   
For  five  countries,  only  data  grouped  by  the  shortlist  in  Table  4.3  were  used,  either  because  too  few 
years’ (Brunei Darussalam and Kazakhstan) or no data (Russian Federation, Ukraine and Belarus) were 
reported by ICD code. Shortlist categories were expanded by using the cause‐fraction distribution within 
each shortlist category by year, age, sex from the GBD2013 study results (4‐6). For Russia, Belarus and 
Ukraine, HIV deaths recorded in the death registration data were substantially miscoded to tuberculosis 
(cause 30), lower respiratory infections  (390), other infectious diseases  (370), lymphomas and multiple 
myeloma  (760),  other  malignant  neoplasms  (780),  and  endocrine,  blood  and  immune  disorders  (810). 
Deaths  in  these  categories  falling  in  the  characteristic  HIV  age  pattern  were  recoded  to  HIV  (100), 
according to the age‐sex‐specific HIV mortality estimates from UNAIDS (Section 8.2).  

World Health Organization Page 20


 
4.5 Adjustment of specific causes
Estimates for tuberculosis deaths were compared with the WHO estimates (also based on an analysis of 
death registration data and surveillance data) and where the death registration numbers were lower, an 
average  of  the  two  sets  of  estimates  was  used.  This  affected  mainly  small  countries  (Antigua  and 
Barbuda, Barbados, Grenada, Iceland, Kuwait, and Luxembourg. 
Estimates  for  HIV  deaths  were  compared  with  UNAIDS/WHO  estimates.  For  seven  mainly  small 
countries, an average of the two sets of estimates were used:  Bahamas, Barbados, Guatemala, Jamaica, 
Saint Lucia, The Former Yugoslav Republic of Macedonia, and Saint Vincent and the Grenadines. 
Estimates  for  malaria  deaths  were  compared  with  WHO  estimates  (see  Section  8.3)  and  replaced  by 
WHO estimates for seven countries where the WHO estimates summed across all years were lower than 
those  from  the  death  registration  data.  This  affected  malaria  deaths  for  Brazil,  Columbia  ,  Ecuador, 
Guatemala, the Republic of Korea, Panama and the Philippines. 
WHO  estimates  for  maternal  deaths  include  an  upwards  adjustment  for  under‐recording  of  maternal 
deaths  in  death  registration  data  (16).  Maternal  deaths  were  adjusted  using  these  country‐specific 
factors, and all other causes adjusted pro‐rata. 
An adjustment was made for estimates of deaths due to cancer of the colon and rectum for Australia.  In 
Australia, the term "bowel cancer" is often used as a synonym for large intestine on death certificates 
(17). However, as the bowel does not refer to a specific site in the digestive tract, the ICD‐10 directs the 
coding of the term "bowel cancer" to C260. The GHE grouping for colon and rectum cancers is C18‐C21. 
As many codes in C260 are a cancer of the colon or rectum, there will be an under estimate in this GHE 
grouping, as C26 is included in "other malignant neoplasms". For Australia, deaths coded to C260 were 
included in the GHE cause category 650 for colon and rectal cancers. 
Relatively small numbers of deaths coded to depression in some countries were re‐assigned to suicide. 
Deaths due to alcohol and drug use disorders include alcohol and drug poisoning deaths coded to the 
injury  chapter  of  ICD  (see  Annex  Table  A).  These  were  adjusted  as  described  in  Section  8.14  to  re‐
allocate unspecified drug dependence, multiple drug use, and unspecified poisoning.  
Where necessary, road injury deaths were adjusted upwards to take account of additional surveillance 
data provided by countries (see Section 8.15). Homicide deaths were similarly adjusted where relevant 
to take account of homicide data from the police/justice sector (see Section 8.16) 
Estimates of  deaths due  to  conflicts (see  Section 8.17)  were  compared with  estimates from  the  death 
registration data year by year and added “outside‐the‐envelope” for country‐years where they are not 
included in death registration data. 
Death rates for some specific conditions were extreme outliers in a few countries. These outliers were 
adjusted as follows: 
 Death rates for skin disease for Barbados were replaced by the average of rates for Antigua and 
Barbuda, Trinidad and Tobago, Saint Lucia and Saint Vincent and the Grenadines, 
 Death rates for otitis media for Saint Vincent and the Grenadines were replaced by the average 
of rates for Antigua and Barbuda, Trinidad and Tobago, and Saint Lucia, 
 Death rates for eating disorders for Guatemala were adjusted using rates for Mexico 
 Death  rates  for    upper  respiratory  tract  infections  for  Kyrgystan  were  replaced  by  estimates 
based on those in GBD2015, which were somewhat higher than the rates for Tajikistan 

World Health Organization Page 21


 
 Death  rates  for    upper  respiratory  tract  infections  for  Uzbekistan  were  replaced  by  estimates 
based on the rates for Kazakhstan 
 Death  rates  for  neurological  disorders  in  Kazakhstan  were  reduced  by  75%  to  bring  them  into 
line with other countries in the region. 
 
5 Causes of death for children under age 5 years
5.1 Child deaths
Cause‐specific  estimates  of  deaths  for  children  under  age  5  were  estimated  for  15  cause  categories 
using methods described elsewhere by Liu et al. (18) and a companion technical paper in this series (19). 
Previously published estimates for years 2000‐2013 (20) have been updated to take account of revisions 
in child mortality levels (10), as well as cause‐specific estimates as described in Section 8. Inputs to the 
multivariate cause composition models are also updated.  The specific methods used for child causes of 
death for China and India are described in Sections 5.2 and 5.3. 
The fifteen cause categories used for the WHO‐MCEE estimates of under 5 deaths for years 2000‐2015 
(see  Annex  Table  E)  include  all  the  major  causes  of  neonatal  (0‐27  days),  postneonatal  (1‐59  months) 
and 1‐4 year deaths and two residual categories containing all remaining causes of death. These residual 
categories  (“Other  Group  1”  and  “Other  Group  2”).  Cause  groups  such  as  “Congenital  malformations” 
and “Injuries” were expanded to the full GHE cause list (Annex Table A) for neonatal and under 5 deaths 
using sub‐cause distributions derived from the GBD2015 estimates (4‐6).   
Note that the WHO‐MCEE cause estimates and the GBD2105 sub‐cause distributions are derived from 
death  registration  data  for  those  countries  with  useable  death  registration  data.  Modest  differences 
between  WHO‐MCEE  estimates  published  in  February  2016  and  these  GHE2015  estimates  are  mainly 
due to new estimates for deaths due to measles and crises, as well as some revisions to malaria deaths 
in low‐burden countries. 

5.2 Child deaths in China


Estimates of causes of death under age 5 were based on a separate analysis of the China  Maternal  and  
Child  Surveillance System (MCMSS). Cause‐specific estimates of deaths for children under age 5 were 
estimated  for  15  cause  categories  using  data  obtained  from  China    Maternal    and    Child    Surveillance 
System  (MCMSS)  for  years  2000‐2015  by  age‐sex‐residency‐region  strata.  The  methods  used  are 
described in more detail in a technical paper in this series (19). The fifteen cause categories used for the 
WHO‐MCEE estimates of under 5 deaths for years 2000‐2015 (see Annex Table E) include all the major 
causes  of  neonatal  deaths  (0‐27  days),  and  deaths  at  ages  1‐59  months  and  two  residual  categories 
containing all remaining causes of death. These residual categories (“Other Group 1” and “Other Group 
2”) and cause groups such as “Congenital malformations” and “Injuries” were expanded to the full GHE 
cause list (Annex Table A) for neonatal and  under 5 deaths using cause distributions derived from  the 
GBD2015 estimates for child causes. 
Total  number  of  deaths  were  estimated  based  on  subnational  live  births  and  MCMSS  strata‐
specific  mortality  rates  smoothed  using  a  three‐year  moving  average,  and  normalized  to  fit  IGME  
all‐cause  number  of  death  estimates.  Cause‐specific  death  proportions  from  MCMSS,  smoothed  
using  a  7‐year  moving  average,  were  applied  to  the  estimated  total  number  of  deaths  to  obtain  
the  estimated  number  of  deaths  by  cause by strata prior to summing to obtain national estimates.  

World Health Organization Page 22


 
5.3 Child deaths in India
In order to estimate trends in under 5 causes of death for India, the previously developed subnational 
analyses  were  further  refined  and  used  to  develop  national  estimates  for  years  2000‐2015.  For 
neonates,  a  verbal  autopsy  multi‐cause  model  (VAMCM)  based  on  37  sub‐national  Indian  community‐
based VA studies was used to predict the cause distribution of deaths at state level. The resulting cause‐
specific  proportions  were  applied  to  the  estimated  total  number  of  neonatal  deaths  to  obtain  the 
estimated number of deaths by cause at state level prior to summing to obtain national estimates.  
For children who died in the ages of 1‐59 months in India, the previously developed multi‐cause model 
(21)  was  rerun  for  years  2000‐2015  after  an  updated  systematic  review  was  conducted  to  identify  27 
new study data points of sub‐national community‐based VA studies, plus 22 sets of observations for the 
Indian states derived from the Million Death Study (22). Nine cause categories were specified, including 
measles  plus  the  eight  specified  in  the  post‐neonatal  VAMCM  for  other  countries.  State‐level  measles 
deaths  were  then  normalized  to  fit  the  national  measles  estimates  produced  by  the  WHO  IVB.  State‐
level AIDS and malaria estimates were provided by UNAIDS and WHO malaria program, respectively. All 
cause fractions were adjusted to sum to one. The state‐level estimates were collapsed to obtain national 
estimates at the end.  
 

World Health Organization Page 23


 
6 Causes of death for China 2000‐2015
6.1 Data sources for causes of death
Cause‐specific mortality data for China  were available from three sources – the sample vital registration 
(VR)  system  data  for  years  1987  to  2012  (23),  summary  deaths  tabulations  from  the  Diseases 
Surveillance Points (DSP) system for years 1995‐1998 and 2004‐2012 (24, 25) and the newly merged and 
expanded  VR  and  DSP  system  for  2013,  referred  to  as  the  Death  Registration  (DR)  system  (26).  The 
Death  Registration  system  also  includes  larger  numbers  of  in‐hospital  deaths  so  that  the  total  deaths 
recorded in the system reached 4 million deaths in 2012 (27). The numbers of deaths recorded in the 
sample representative sites for DSP, VR and DR systems is summarized in Table 6.1 below. 
 

Table 6.1 Total deaths and population covered by the Chinese vital registration system (VR) , the Disease 
Surveillance Points system (DSP) and the newly merged Death Registration system (DR). 
 
Number of deaths  Population 
Year  VR  DSP  DR  VR  DSP  DR 

2000  711,946  …  …  117,183,678  …  … 


2001  …  …  …  …  …  … 
2002  …  …  …  …  …  … 
2003  626,392  …  …  102,889,945  …  … 
2004  295,906  430,994  …  55,288,841  71,173,205  … 
2005  310,826  437,490  …  57,272,144  71,487,277  … 
2006  379,057  347,057  …  72,240,261  66,012,299  … 
2007  475,289  401,008  …  79,101,646  71,476,477  … 
2008  471,219  424,683  …  …  73,928,499  … 
2009  505,021  437,550  …  …  75,020,489  … 
2010  558,915  453,211  …  90,158,748  78,766,626  … 
2011  775,458  437,490  …  124,960,668  77,396,478  … 
2012  929,249  459,836  …  147,969,227  77,215,997  … 
2013  …  …  1,463,851  …  …  227,236,284 
             
… data not available. 

 
   

World Health Organization Page 24


 
These  sets  of  data  were  assessed  and  compared  for  suitability  in  estimating  2000‐2015  cause‐specific 
mortality for China at the national level. The VR and DSP datasets gave quite similar cause distributions 
at  major  cause  group  level  by  age,  across  the  period  2000‐2010.  Additionally,  comparison  for  more 
detailed major causes of death did not give any clear indication that one data set was of systematically 
higher quality than the other. We therefore based the previous 2000‐2012 cause of death estimates for 
China on an average of the estimates from the two systems (1).  
With the merger of the two systems in 2013, and the expansion of urban sample sites, the urban‐rural 
composition  of  the  sampled  populations  changed  to  be  more  nationally  representative.  For  earlier 
years,  WHO  analyses  had  re‐weighted  urban  and  rural  samples  from  DSP  and  VR  to  give  approximate 
national  representativeness.    However,  the  DR  dataset  for  2013  also  uses  a  different  set  of  cause 
categories,  not  entirely  consistent  with  the  earlier  datasets.    We  mapped  cause  categories  from  the 
three  datasets  to  GHE  cause  categories  and  examined  the  resulting  cause‐specific  time  trends.  There 
were inconsistencies between the DSP+VR based results and the 2013 results which were not resolvable 
given the available cause‐specific information.  

6.2 Estimation of deaths by cause for ages 5 and over


We  also  compared  these  results  with  the  GBD2015  national  cause‐specific  trends  for  China  (4‐6)  and 
found reasonable consistency for the 2015 results for most but not all causes. For causes for which WHO 
has  specific  estimates  as  described  in  Section  8  below,  these  estimates  were  used.  For  other  causes, 
cause  fractions  from  the  GBD2015  estimates  were  used,  adjusted  to  the  WHO  envelope  for  these 
causes. The GBD2015 estimates were derived from available Chinese data on causes of death at national 
and sub‐national levels, with major inputs coming from the DSP and VR sample systems for years 2000‐
2012, with additional data on deaths in Chinese hospitals (27).  
The DR 2013 data were used to make the following adjustments to WHO and GBD2015 inputs: 

(1) 2011‐2015 estimates for road injury deaths were revised based on analysis of the DR 
2013 data.  The estimated total road injury deaths for China in 2013 was revised upwards 
from 261,000 to 272,000, which is still somewhat lower than the 310,000 estimated by 
GBD2015.   
(2) The following  GBD2015 cause fractions were adjusted based on comparisons with 2013 
data: 
a. Diabetes 17% increase 
b. Epilepsy  17% increase 
c. Other neurological 53% increase 
d. COPD increase in 2000 of 6%, dropping to 0% in 2010 and beyond (based on 
trend in VR/DSP data for 2000‐2010 
e. Other respiratory: increase in 2000 of 20%, dropping to 0% in 2010 and beyond 
(based on trend in VR/DSP data for 2000‐2010 
f. Suicide: approximately 5% increase across all years 

Other adjustments for specific causes are described in Section 8 below.  
 
 

World Health Organization Page 25


 
6.3 Comparison of GHE estimates with death registration data and GBD2015
GBD2015  results  were  used  as  inputs  to  estimate  cause  fractions  by  country,  age,  sex  and  year  for 
causes  of  death  at  ages  five  years  and  above  for  which  death  registration  data  and/or  WHO  and  UN 
Interagency analyses were not available. For this set of causes, GBD2015 national  estimates for death 
rates  at  ages  5  and  over  for  China  in  years  2000‐2015  were  used.  For  each  year  2000  to  2015,  cause 
fraction  distributions  were  then  computed  for  the  set  of  causes  excluding  WHO/Interagency  cause‐
specific  estimates.  For  countries  where  these  cause  fractions  were  used,  they  were  applied  to  the 
country‐level  residual  mortality  envelopes  by  age  and  sex  after  the  WHO/Interagency  cause‐specific 
estimates were subtracted from the WHO all‐cause envelopes.  
Table 6.2 summarizes and compares the estimates for total deaths in 2015 for selected causes from the 
DR 2013 (scaled to the China all‐cause envelope), from GHE2015 and from GBD2015. 
 

Table  6.2  Estimated  total  deaths  (‘000s)  for  China  in  2013,  selected  causes  from 
DR2013 (rescaled to total deaths by age and sex), GHE2015 and GBD2015. 
Cause  DR2013  GHE2015  GBD2015 
Tuberculosis  32.2  33.6  48.9  
HIV  6.4  23.1   41.8  
Diarrheal diseases  4.6  9.2  5.9  
Malaria  0.0  0.0   0.2  
Lower respiratory infections  215.9  217.1   205.1  
other infectious diseases  117.6  53.4   t  
Maternal causes  3.1  4.8  2.9  

Neonatal causes  81.6  68.8  75.4  


Cancers  2,290.3  2,309.6  2,343.8  
      Oesophagus cancer  242.9  196.6  202.0  
      Stomach cancer  265.0  326.5  334.7  
      Lung cancer  455.3  635.6  580.0  
      breast cancer  30.2  49.0  69.2  
Cardiovascular diseases  4,105.3  4224.0  3,829.6  
Chronic respiratory diseases  1,084.3  1,061.2   957.8  
Other noncommunicable diseases 1,058.2  1,196.5  8221.6  
Road injury  281.5  268.1  310.5  
Poisoning  34.6  22.4  22.3  
Suicide  135.2  138.6  135.7 
Homicide  9.9  11.9  20.0  
Other injury  289.2  255.3  777.2  
  Source for GBD2015 (4‐6) 

   

World Health Organization Page 26


 
7 Causes of death for India 2000‐2015
7.1 Sample Registration System data
Analysis of causes of death for India was based on data from the Sample Registration system (SRS) for 
the periods 2001‐2003 (28, 29) and 2010‐2013 (30, 31).  These data were derived from representative 
samples  of  deaths  in  the  SRS  sampling  areas,  for  which  verbal  autopsy  methods  were  used  to  assign 
cause of death.  The Sample Registration System monitors a representative sample population of over 6 
million  people  in  over  1  million  homes  in  India.  In  2013,  a  total  of  7,597  sample  units  covered  a  total 
population of 7.5 million people, of whom 2.0 million were in urban areas and 5.5 million in rural areas.  
In  2001  the  Indian  Registrar  General  Surveyor  introduced  an  enhanced  form  of  verbal  autopsy  for 
assessing  the  cause  of  death.  Verbal  autopsy  is  a  method  of  ascertaining  the  cause  of  death  by 
interviewing a family member or caretaker of the deceased to obtain information on the clinical signs, 
symptoms and general circumstances that preceded the death. Details of methods and validation have 
been  reported  elsewhere  (29,  30).  Verbal  autopsy  reports  were  independently  coded  to  ICD‐10 
categories by at least two of a total of 130 physicians trained in ICD‐10 coding. In case of disagreement 
on the ICD‐10 codes at the chapter level, reconciliation between reports was conducted, followed by a 
third senior physician’s adjudication.  
A total of 122,848 deaths between January 2001 and December 2003, and a total of 182,827 deaths for 
2010‐2013 were assigned causes of death by verbal autopsy. Verbal autopsies could not be conducted 
for around 10% of the deaths for reasons such as family migration or change of residence.  
The cause‐specific proportion of deaths in each five‐year age category from 0 to 79 years and for people 
aged 80  years  and  over  was  weighted  by  the inverse  probability  of  a  household  being  selected  within 
rural  and  urban  subdivisions  of  each  state  to  account  for  the  sampling  design.  National  estimates  for 
deaths and mortality rates were based on reweighted urban and rural estimates for India, by age, sex 
and area.  
The GHE analysis is based on the resulting national‐level cause‐specific mortality proportions derived for 
GHE cause categories from the SRS data. GBD2015 cause fractions were used to redistribute deaths to 
detailed sub‐cause categories in cases where the SRS cause categories were broader than the GHE cause 
categories. 

7.2 Comparison of GHE estimates with SRS and GBD2015


For  causes  for  which  full  time  series  estimates  for  years  2000‐2015  were  not  available  from  WHO 
technical programs and UNAIDS (see Section 8), the trends for the full period 2000‐2015 were estimated 
as  follows.  We  made  use  of  the  trends  estimated  by  IHME  in  the  GBD2015  study  (2).  The  India  data 
sources used by IHME can be inspected on their website (12). The GBD2015 estimates for years 2000‐
2015 were rescaled for consistency with the total deaths across all such causes estimated from WHO life 
tables  and  cause‐specific  estimates.  Age‐sex‐cause  specific  ratios  of  SRS‐based  deaths  to  rescaled 
GBD2015  deaths  were  calculated  from  the  SRS  data  for  period  2002  (2001‐2003)  and  2011.5  (2010‐
2013). The scale factors were linearly interpolated for years 2003‐2011 and extrapolated to year 2000 
and  2015.  They  were  then  applied  to  the  GBD2015  estimates  to  generate  full  time  series  for  these 
causes consistent with the WHO analyses of the SRS data for 2001‐2003 and 2010‐2013. The remaining 
cause‐specific estimates were based on information from WHO technical programmes and UNAIDS on 
specific causes as described in Section 8.  
 

World Health Organization Page 27


 
8 Methods for specific causes with additional information
8.1 Tuberculosis
For countries with death registration data, tuberculosis mortality estimates were generally based on the 
most  recently  available  vital  registration  data.  For  other  countries,  total  tuberculosis  deaths  were 
derived  from  latest  published  WHO  estimates  (32),  together  with  more  detailed  unpublished  age 
distributions based on the  VR data and notifications data.  

8.2 HIV/AIDS and sexually transmitted diseases


(a) High HIV countries
For 43 countries with significant HIV epidemics, explicit efforts were made to ensure consistency of all‐
cause and HIV mortality estimates across the period 2000‐2015 in the 2016 revision of WHO life tables 
and  all‐cause  mortality  “envelopes”  (11).  These  countries  are  identified  in  Annex  Table  D.  For  these 
countries, HIV mortality estimates from either UN Population Division (8) or UNAIDS (33) were revised 
using updated Spectrum models for 1985‐2015. These models took into account the WPP 2015 revisions 
to demographic data and all‐cause mortality, as well as 2015 UNAIDS files  with a range of 2016 updates 
to  the  Spectrum/AIM  software  including  new  patterns  of  adult  mortality  on  ART  and  age  at  ART 
initiation among pediatric patients and the re‐fitting of all the EPP curves (11, 34). 

(b) Countries with useable vital registration data


For  countries  with  death  registration  data,  HIV/AIDS  mortality  estimates  were  generally  based  on  the 
most  recently  available  vital  registration  data  except  where  there  was  evidence  of  misclassification  of 
HIV/AIDS  deaths.  In  such  cases,  a  time  series  analysis  of  causes  where  there  was  likely  misclassified 
HIV/AIDS deaths was carried out to identify and re‐assign such deaths.  

(c) Other countries


For other countries, estimates were based on UNAIDS estimated HIV/AIDS mortality (33). UNAIDS does 
not  estimate  HIV  deaths  for  the  following  countries:  Comoros,  Libya,  Micronesia,  Samoa,  Seychelles, 
Solomon Islands, Tonga and Vanuatu. HIV estimates for these countries were based on previous WHO 
GHE2013  estimates  with  projections.    It  was  assumed  based  on  advice  from  UNAIDS  that  1%  of  HIV 
deaths under age 5 occurred in the neonatal period. 

(d) South Africa


For  development  of  previous  WHO  life  tables,  South  Africa  was  classified  as  “high  HIV”  and  explicit 
efforts  made  to  ensure  consistency  of  all‐cause  and  HIV  mortality  estimates.  For  the  WPP  2015,  UN 
Population Division used Spectrum (34) with input assumptions consistent with those of UNAIDS in mid‐
2014 to model all‐cause mortality for South Africa. This resulted in estimates for HIV deaths that were 
not consistent with the most recent estimates by UNAIDS (33). Figure 8.1 shows total HIV deaths by year 
as  estimated  by  UN  Population  Division  for  the  WPP2015  life  tables,  UNAIDS  revision  in  2016,  IHME 
estimates  from  GBD2013,  and  current  WHO  draft  estimates  along  with  the  WHO  previous  GHE2013 
estimates for 2000‐2012 (1).  

World Health Organization Page 28


 
Figure 8.1.  Comparison of estimates of total HIV deaths 2000‐2015, South Africa 

 
 
During the 2016 revision of WHO life tables drawing on inputs from WPP2015, it was realized that use of 
either the UNPD or UNAIDS estimates of HIV mortality would either result in implausibly low levels of 
non‐HIV  mortality,  or  high  levels  of  all‐cause  mortality  not  consistent  with  available  estimates  of 
completeness of death registration data. Figure 8.2 compares estimates of non‐HIV mortality and total 
all‐cause  mortality  from  the  various  agencies,  and  Figure  8.3  compares  the  implied  completeness 
estimates  of  South  African  death  registration  data  (calculated  by  dividing  total  registered  deaths  by 
estimated total deaths from all causes). 

Figure 8.2.  Comparison of estimates of total non‐HIV deaths and all‐cause deaths, 2000‐2015, South Africa 

World Health Organization Page 29


 
Figure 8.3.  Comparison of estimates of completeness of death registration, 2000‐2014, South Africa 

 
 
Given  the  inconsistencies  between  the  HIV  mortality  estimates,  all‐cause  mortality  estimates  and 
implied completeness of death registration, we decided to maximise consistency with the previous WHO 
estimates of HIV and non‐HIV mortality for years 2000‐2012 (1) released in 2014. Estimates of both non‐
HIV and HIV mortality were adjusted upwards for earlier years near 2000, as the implied completeness 
of  death  registration  was  too  high  for  the  previous  WHO  estimates.  Non‐HIV  mortality  rates  were 
projected forwards drawing on the trend estimates from UNPD and HIV mortality rates were projected 
forwards based on the trends in the UNAIDS estimates for HIV deaths.  

8.3 Malaria
Under 5 deaths in countries with high quality VR data 
For countries in which death reporting is estimated to capture > 50% of all deaths and a high proportion 
of malaria cases are parasitologically confirmed, reported malaria deaths are adjusted for completeness 
of death reporting. For countries in elimination programme phase, reported malaria deaths are adjusted 
for completeness of case reporting. 

Under 5 deaths in countries outside the WHO African Region and low transmission countries in Africa 
For countries (i) outside the African Region in which death reporting is estimated to capture ≤ 50% of all 
deaths  or  a  high  proportion  of  malaria  cases  are  not  parasitologically  confirmed,  or  (ii)  in  the  African 
Region  where  estimates  of  case  incidence  were  derived  from  routine  reporting  systems  and  where 
malaria comprises less than 5% of all deaths in children under 5,1 case fatality rates are used to derive 

                                                            
1
Algeria, Botswana, Cape Verde, Comoros, Eritrea, Ethiopia, Madagascar, Namibia, Sao Tome and Principe, South
Africa, Swaziland, and Zimbabwe

World Health Organization Page 30


 
number of deaths from case estimates. A case fatality rate of 0∙256% is applied to the estimated number 
of  P.  falciparum  cases,  being  the  average  of  case  fatality  rates  reported  in  the  literature  (35‐37)  and 
unpublished  data  from  Indonesia,  2004‐2009  (correspondence  with  Dr.  Ric  Price,  Menzies  School  of 
Health Research).  A case fatality rate of 0.0375% is applied to the estimated number of P. vivax cases, 
representing  the  mid‐point  of  the  range  of  reported  case  fatality  rates  (38).  The  number  of  cases 
reported by a Ministry of Health is adjusted to take into account (i) incompleteness in reporting systems 
(ii) patients seeking treatment in the private sector, self‐medicating or not seeking treatment at all, and 
(iii) potential over‐diagnosis through the lack of laboratory confirmation of cases. 

Under 5 deaths in South Sudan and high transmission countries in the WHO African Region.   
For countries in the African Region where malaria comprises 5% or more of all deaths in children under 
5, malaria deaths were estimated using a multinomial logistic regression model fitted to available verbal 
autopsy data sets. This model is described in more detail elsewhere and draws on geospatial estimates 
of  parasite  prevalence  rates  produced  by  the  Malaria  Atlas  Project  at  Oxford  University  in  close 
collaboration with WHO (39). 

Malaria deaths at ages 5 and over.   
The  estimated  malaria  mortality  rate  in  children  under  5  years  for  a  country  was  used  to  determine 
malaria transmission intensity and the corresponding malaria‐specific mortality rates in older age groups 
(53). 
These estimates for malaria deaths 2000‐2015 are consistent with those published in the World Malaria 
Report  2016  (39),  except  for  some  slight  differences  resulting  mainly  from  malaria  deaths  coded  in 
death registration data for some countries. 

8.4 Whooping cough


Recognizing  the  limited  data  to  support  modelling  of  pertussis  mortality,  the  World  Health 
Organization’s  Department  of  Immunization  Vaccines  and  Biologicals’  Quantitative  Immunization  and 
Vaccines  Related  Research  (QUIVER),  recommended  in  2009  that  a  revised  pertussis  model  be 
developed to specifically address uncertainty in the model inputs and parameter values.  Inputs to the 
current  model  are  country‐  and  year‐specific  estimates  of  population  by  single  year  of  age  and 
estimated  pertussis  immunization  coverage  (40).  Age‐,  country‐,  and  immunization  history‐  specific 
estimates  of  the  probability  of  initial  infection,  probability  that  an  infected  individual  develops  typical 
symptoms of a case of pertussis and the probability that a case of pertussis will die were estimated using 
structured  expert  judgment  (41,  42).  Annual  deaths  attributable  to  pertussis  infection  during  the 
neonatal period (5% of estimated pertussis deaths 0‐11 months of age), from age 1‐11 months of age 
(estimated  as  95%  of  deaths  1‐11  months  of  age)  and  12‐59  months  of  age  were  estimated  for  each 
country  for  the  years  2000  –  2012.  The  pertussis  cause  fraction  was  assumed  to  be  constant  to 
extrapolate forwards to 2015. Pertussis deaths at ages 5 and over were estimated from useable death 
registration data or GBD2015 analyses. 

8.5 Measles
To  estimate  deaths  attributable  to  measles,  a  new  model  of  measles  mortality  developed  by  WHO 
Department  of  Immunization,  Vaccines  and  Biologicals  (IVB)  was  used  to  first  estimate  country‐and‐
year‐specific  cases  using  surveillance  data  (43,  44).  The  improved  statistical  model  firstly  estimates 
measles  cases  by  country  and  year  using  surveillance  data  and  making  explicit  projections  about 
dynamic transitions over time as well as overall patterns in incidence.  

World Health Organization Page 31


 
The cases are then stratified by age classes based on a model fitted to data stratified by national GDP 
and vaccine coverage. The results are applied to age‐specific case fatality ratios for each country (45‐47) 
and  then  aggregated  again  to  produce  overall  measles  deaths.  The  estimates  used  here  are  from  an 
update  to  take  into  account  trends  in  case  notifications  and  vaccine  coverage  up  to  and  including  the 
year 2015 (48).  
Measles  deaths  at  ages  5  and  over  were  estimated  from  useable  death  registration  data  or  GBD2015 
analyses. 

8.6 Hepatitis‐attributable deaths


For liver cancer and cirrhosis of the liver, the GBD2015 estimated deaths for four aetiological categories: 
hepatitis B virus (HBV) infection, hepatitis C virus (HCV) infection, alcohol, and “other causes”. DisMod‐
MR  2.1  was  used  to  model  the  proportions  of  liver  cancers  and  liver  cirrhosis  due  to  these  four  sub‐
causes  using  data  derived  from  systematic  reviews  of  literature  on  the  aetiology  of  liver  cancers  and 
liver cirrhosis (5). Relevant covariates used in DisMod‐MR 2.1 were apparent alcohol consumption (litres 
per  capita),  hepatitis  B  surface  antigen  (HBsAg)  seroprevalence,  and  hepatitis  C  (anti‐HCV  IgG) 
seroprevalence,  and  a  binary  covariate  indicating  countries  with  a  predominantly  Muslim  population 
(associated with low alcohol consumption).  
To  ensure  coherent  results  between  the  cirrhosis  and  the  liver  cancer  etiologies,  the  results  from  the 
liver  cancer  etiology  models  were  transformed  into  covariates  that  were  then  used  in  the  cirrhosis 
etiology  models.  The  results  from  the  cirrhosis  etiology  models  were  then  used  in  the  liver  cancer 
proportion models. The DisMod proportions for the underlying liver cancer and cirrhosis etiologies were 
then  squeezed  to  100%  and  these  final  proportions  were  applied  to  the  total  liver  cancer  and  liver 
cirrhosis  to derive the estimates for the four etiologies.  
IARC  has  also  carried  out  an  analysis  of  the  hepatitis  B  and  hepatitis  C  fractions  of  total  liver  cancer 
cases.  Estimates  for  50  countries  have  been  published  (49)  and  regional  and  global  estimates  are  in 
preparation (50). Since there is considerable time lag between hepatitis infection and death from liver 
cancer,  the  proportions  attributable  to  HBV  and  HCV  infection  relate  to  hepatitis  seroprevalence 
distributions in the past, when hepatitis C was less prevalent than in recent years. The time series used 
in the IARC paper vary from country to country depending on available data, but typically contain data 
ranging from the early 1990s to the early 2000s, in some cases out to 2010. The data for China are for 
the  range  1954‐2010.  Details  of  the  time  periods  for  the  data  used  in  the  IHME  analyses  are  not 
available. On the other hand, the IHME analyses included a complete set of subcause categories as they 
are  also  estimating  alcohol  and  other  causes  as  well  as  hepatitis  infection  (ensuring  that  all  cause 
fractions  add  to  100%).  The  IARC  analyses  address  only  hepatitis  B  and  C  with  the  potential  for  over‐
estimation of causal fractions.   
Table  8.1  summarizes  the  global  attributable  fractions  for  HBV  and  HCV  caused  liver  cancer  and  liver 
cirrhosis from WHO, IARC  and IHME, together  with the GHE2015 final  estimates, derived as  described 
below. In estimating the sub‐causes of liver cancer and cirrhosis for GHE2015, we drew on the GBD2015, 
GBD2013 and IARC analyses as follows. The GBD2015 cause fractions for liver cancer were revised pro‐
rata  to  adjust  the  HBC  fraction  of  HBV+HCV  caused  liver  cancer  by  country/region  group  to  the  IARC 
estimates  shown  in  Table  8.2.  We  also  revised  the  “other”  category  downwards  to  the  proportions 
estimated in GBD2013, shifting the excess deaths to HBV and HCV. This resulted in an overall estimate of 
the fraction of liver cancer attributable to hepatitis (HBV or HCV) similar to the IARC estimates. Cirrhosis 
death  attributions  were  similarly  adjusted  drawing  on  the  HCV/HBV  proportions  estimated  for  liver 
cancer.  

World Health Organization Page 32


 
Table 8.1 Global attributable fractions for liver cancer and cirrhosis, latest year, WHO, IARC and GHE. 
  WHO (51)  IARC (49,50) GBD2013 GBD2015 GHE2015 
Liver cancer     
    HBV  53  53 37 33 43 
    HCC  25  20 42 21 17 
    Alcohol    11 30 30 
    Other    10 16 10 
Cirrhosis     
    HBV  30  26 29 40 
    HCC  27  29 25 22 
    Alcohol    32 27 25 
    Other    13 19 13 
HBV+HCV percent of     
   Liver ca  78  73 79 53 60 
   Cirrhosis  57  55 54 62 

 
 

Table  8.2  Fractions  of  liver  cancer  caused  by  HBV  and  HPC  estimated  by  IARC  (49,  50)  and  IHME  (4). 
IHME GBD2015 results shown are for year 2015. 
    HBV %of (HBV+HVC) (HBV+HCV) % 
Region    IARC GBD2015 IARC GBD2015 
Eastern Asia    42 57 89 58 
Latin America    56 69 64 54 
China    7 18 74 49 
India    24 29 69 63 
Russia    29 61 51 39 
Northern Africa    85 81 88 75 
Northern America    73 76 45 40 
Northern Europe    78 69 36 54 
Oceania    29 62 70 51 
Rest of Europe    71 76 62 54 
Sub‐Saharan Africa    30 29 71 57 
Western‐Central Asia    47 57 74 67 
World    73 62 73 53 
 
 

World Health Organization Page 33


 
8.7 Schistosomiasis
For the WHO update of burden of disease for year 2004 (52), the incidence and prevalence of cases of 
schistosomiasis  infection  were  separately  estimated  by  country  for  S. mansoni,  S. haematobium  and 
S. japonicum plus S. mekongi.  The GBD 2004 estimated that schistosomiasis was responsible for around 
41 000 deaths globally (excluding attributable cancer deaths) and 36 000 in sub‐Saharan Africa, although 
others  have  argued  that  the  figure  should  be  much  higher  (53).  Van  der  Werf  et  al  (54),  using  limited 
data  from  Africa,  estimated  that  schistosomiasis  caused  210 000  deaths  annually.  For  the  GBD  2004 
update,  very  limited  available  data  was  used  to  conservatively  estimate  annual  case  fatality  rates  for 
prevalent  cases  at  0.01%  for  S. mansoni,  0.02%  for  S. haematobium,  and  0.03%  for  S. japonicum  and 
S. mekongi. There were estimated to be 261 million prevalent cases of schistosomiasis infection in 2004. 
The  GBD2015  study  estimated  that  there  252  million  prevalent  cases  of  schistosomiasis  infection  in 
2015, and 4,365 deaths due to schistosomias, giving an implied average case fatality rate of 0.002%, an 
order of magnitude lower than earlier WHO estimates. The GBD2015 implied case fatality rates for the 
Middle  East  and  North  Africa,  for  Latin  America,  and  for  Southeast  Asia,  East  Asia  and  the  Pacific  are 
0.009%,  0.017%  and  0.08%  respectively.  These  are  substantially  higher  than  the  implied  African  case 
fatality  rate  of  0.001%.  Revised  case  fatality  rates  of  0.0075%  for  S. mansoni,  0.015%  for 
S. haematobium were applied to the prevalence rates estimated by GBD2015 (6) to revise the estimates 
of schistosomiasis deaths for GHE. This resulted in an estimate of 21,170 deaths in sub‐Saharan Africa 
and 24,067 deaths globally in 2015. 

8.8 Cycsticercosis, echinococcosis and food‐borne trematodes


In 2007, the World Health Organization (WHO) established the Foodborne Disease Burden Epidemiology 
Reference Group (FERG) to estimate global and regional burdens of foodborne disease. Included among 
the  parasitic  foodborne  diseases  analysed  were  cysticercosis,  echinococcis,  and  food‐borne 
trematodosis. In 2015, the FERG published regional and global estimates of deaths and DALYs for these 
diseases  for  the  year  2010  (55,  56).  The  GBD2015  time  series  estimates  of  deaths  for  these  three 
diseases were scaled to match the underlying FERG estimates of deaths by country in 2010.  

8.9 Rabies
GHE2015 estimates of rabies deaths for years 2000‐2015 were updated as follows. Total rabies deaths 
for  years  2011‐2015  were  based  on  more  recent  data  on  reported  human  rabies  deaths  from  the 
Chinese Center for Disease Control and Prevention (57, 58). Rabies deaths for India were revised based 
on the reported deaths in the Indian SRS data for years 2001‐2003 and 2010‐2013 (see Section 7). For 
other countries with more than 10 estimated rabies deaths per year, years 2013 to 2015 were projected 
assuming an average annual rate of decline of 4% based on the trend in the GHD2013 estimates. 

8.10 Leprosy
The GBD2015 estimated that there were 514,200 prevalent cases of leprosy in 2015, with 35% of these 
in India.  Although cause of death data for both India and China contained leprosy deaths, the GBD2015 
estimated zero  deaths  globally.  The implied  case  fatality  rate for  India  of  2% was  applied  to  GBD2015 
estimates of leprosy cases across all countries. Resulting global deaths for leprosy in 2015 were just over 
15,900.  
 
 

World Health Organization Page 34


 
8.11 Ebola
Deaths directly resulting from Ebola virus infection in 2014 and 2015 in Liberia, Sierra Leone and Guinea 
were estimated using the “medium” scenario estimates of Helleringer and Noymer (59). They estimated 
Ebola  deaths  for  three  scenarios  as  follows:  a  “low”  scenario  where  they  consider  that  no  cases  went 
unrecorded,  and  a  high  scenario  where  they  consider  that  there  2.5  times  more  cases  than  recorded. 
The  medium  scenario  considers  70%  more  cases  than  recorded.  There  have  been  new  estimates 
recently  that  suggest  that  there  could  be  up  to  3.5  times  more  cases  than  recorded,  but  these  were 
focused on a small locality, so Helleringer and Noymer did not extend the range all the way to 3.5. 
The Ebola outbreak overwhelmed the healthcare systems of Guinea, Liberia, and Sierra Leone, reducing 
access  to  health  services  for  diagnosis  and  treatment  for  the  major  diseases  that  are  endemic  to  the 
region: malaria, HIV/AIDS, and tuberculosis.  Parpia et al. (60) modelled the impact of reduced access to 
health services on the mortality rates for these three diseases. We took their modelled impact of a 50% 
reduction in treatment coverage to estimate the additional deaths for malaria (under 5), HIV( ages 15 +) 
and tuberculosis (all ages). Their estimates related to March 2014 to March 2015, and we assumed the 
coverage  collapse  would  have  covered  half  a  year  in  2014  and  half  a  year  in  2015.    For  Liberia,  there 
were very few Ebola deaths in 2015 compared to 2014, so we reduced the coverage collapse to 1/3 of 
2015. 
Takashi et al (61) estimated the likely increase in measles deaths resulting from disruption of childhood 
vaccinations during the Ebola outbreak. They projected that after 6 to 18 months of disruptions, a large 
cluster  of  children  unvaccinated  for  measles  would  have  accumulated  across  the  three  countries, 
increasing the expected sizes of regional measles outbreaks and resulting in an additional 5,200 deaths 
(range 2,000 – 16,000).  
Data reported to WHO from the case‐based measles surveillance systems for all three countries to 31 
March 2016 confirmed that there were outbreaks of measles in 2015 in all three countries, although it is 
likely that reported cases do not accurately reflect the magnitude of the outbreaks. We conservatively 
adjusted the 2015 measles deaths to include outbreaks of the same magnitude as those estimated for 
2014 for deaths under age 5, and to increase the 2015 deaths over age 5 so they were 10% higher than 
those in 2014. 
We explored options for estimating other impacts of health system collapse during the Ebola epidemic, 
but decided to limit the estimates to those outlined above, for two reasons.  First, the impact on HIV, TB, 
malaria and measles may be higher because of the direct impact on interventions with a big effect on 
mortality (ART, DOTs, antimalarials, vaccination) and  second, it’s not clear that there would have been 
much pre‐Ebola health system impact on other causes (particularly for adults).   
The  estimated  direct  and  indirect  mortality  impacts  of  the  Ebola  epidemic,  included  in  GHE2015,  are 
summarized in the following Table 8.3. 
 

World Health Organization Page 35


 
 
Table 8.3 Estimated direct and indirect additional deaths associated  
                  with the West African Ebola outbreak of 2014‐2015. 
2014  2015  Under5  Over5  Total 
Guinea 
Ebola  2,635  1,105  324  3,416  3,740 
TB  849  847  336  1,361  1,697 
HIV  339  288  ‐  627  627 
Measles  ‐  550  450  100  550 
Malaria  2,197  2,082  4,199  80  4,279 
Total  6,021  4,872  5,309  5,583  10,892 
Liberia 
Ebola  5,412  155  307  5,260  5,567 
TB  830  560  215  1,175  1,390 
HIV  112  73  ‐  186  186 
Measles  ‐  145  119  25  145 
Malaria  420  253  627  46  673 
Total  6,774  1,186  1,268  6,693  7,961 
SierraLeone 
Ebola  9,025  1,580  952  9,653  10,605 
TB  787  803  324  1,267  1,590 
HIV  106  94  ‐  199  199 
Measles  ‐  4,533  4,532  1  4,533 
Malaria  1,576  1,347  2,844  78  2,923 
Total  11,493  8,357  8,652  11,198  19,851 
Total 
Ebola  17,072  2,840  1,583  18,329  19,912 
TB  2,466  2,211  875  3,802  4,677 
HIV  557  455  ‐  1,012  1,012 
Measles  ‐  5,228  5,101  127  5,228 
Malaria  4,193  3,682  7,670  204  7,875 
Total  24,288  14,415  15,229  23,474  38,703 

 
 

World Health Organization Page 36


 
8.12 Maternal causes of death
Country‐specific estimates for maternal mortality were based on the most recent Interagency estimates 
for years 2000‐2015 (16). A multilevel regression model for the proportion of total female deaths in the 
age  range  15‐49  that  were  due  to  maternal  causes  (PM)  was  developed  using  available  national‐level 
data  from  surveys,  censuses,  surveillance  systems  and  death  registration.  This  regression  model 
included national income per capita, the general fertility rate and the presence of a skilled attendant at 
birth (as a proportion of total births) as covariates to predict trends in maternal mortality. 
 
Because the WHO life tables, and hence the total female deaths in the maternal age range, were revised 
in 2016, the interagency PM estimates have been applied to the new envelopes to estimate numbers of 
maternal  deaths.  This  has  resulted  in  changes  in  the  estimates  of  maternal  deaths  for  some  countries 
with substantial revisions to all‐cause mortality, and small changes to regional and global total maternal 
deaths. 
Note that  the maternal  mortality estimates  include those  HIV deaths  occurring in  pregnant women  or 
within 42 days of end of pregnancy which were considered to be indirect maternal deaths rather than 
incidental. These HIV maternal deaths were subtracted from total HIV deaths as estimated by UNAIDS.

8.13 Cancers
Cause‐specific estimates for cancer deaths in 2012 were derived from Globocan 2012 (62). For countries 
without  useable  death  registration  data,  site‐specific  deaths  were  projected  back  to  year  2000  using 
trend estimates from the GBD2015.   For countries with useable death registration data, cancer deaths 
by  site  were  estimated  from  the  death  registration  data  directly  with  the  various  adjustments  and 
redistributions described in Section 4. 
Karposi sarcoma was excluded from the Globocan estimates as this is almost entirely a manifestation of 
HIV/AIDS, already included in the estimates for HIV/AIDS deaths.  

8.14 Alcohol use and drug use disorders


The  injury  codes  for  accidental  poisoning  by  alcohol  and  by  opioids  are  now  used  to  code  acute 
intoxication  deaths  from  alcohol  and  acute  overdose  deaths  by  opioids.  These  deaths  have  been 
remapped to alcohol use disorders and drug use disorders respectively (see Annex Table A). GBD2015 
estimates for alcohol and drug use disorder deaths, and for accidental poisoning deaths, were used as 
the starting points for the WHO GHE as described in Section 9 below.  
The GBD2013 attributed deaths coded to the ICD‐10 code X49 “Accidental poisoning due to other and 
unspecified chemicals” pro‐rata to the GBD drug‐type cause categories for drug use disorders, based on 
the similarity of the age patterns for unspecified poisoning to those for drug overdose.  This resulted in a 
relatively  large  proportion  of  drug  use  disorder  deaths  for  the  “other  drug  use  disorders”  category. 
Based on a literature review that identified opioid‐dependecy as a large contributor to the deaths in the 
“other  drug  use  disorders”  category,  particularly  where  multiple  drug  use  was  involved  (5,  page  150), 
the GBD2015 redistributed a large  [but undocumented] proportion of the “other drug use” deaths to 
the “opioid use disorder” category.  
An analysis of the age pattern of “Other drug use disorders” for Australia and the USA and comparison 
with the age patterns for “Opioid drug use disorders” and for accidental poisoning by prescription drugs, 
also confirmed that the resulting GBD2015 “Other drug use disorders” should be re‐attributed in part to 
opioid use disorders. Analysis of detailed Australian data for deaths coded to ICD‐10 code F19 “Multiple 

World Health Organization Page 37


 
drug  use,  other  and  unknown  drug  use”  has  shown  that  around  77%  of  these  deaths  involve  opioid 
drugs (63). Based on the adjustment needed to match this result, 24.3% of “other drug use” deaths were 
re‐attributed to opioid use disorders. 
The resulting global deaths for opioid use disorders of 127,373 in 2015 is thus somewhat higher than the 
122,048 estimated by the GBD2015 for the year 2015. The GHE estimates are also more consistent with 
the  CDC  estimates  of  opioid‐caused  deaths  in  the  USA  (64,  65).  Note  that  these  are  deaths  directly 
caused  by  opioid  use.    The  total  attributable  deaths  for  opioid  use  are  much  higher  as  they  include 
deaths due to infectious diseases transmitted via re‐use of injecting equipment, as well as deaths due to 
road injury and suicide. The UNODC’s World Drug Report 2015 estimated there were 187,100 ( 98,300‐
231,400)  drug‐related  deaths  globally  in  2013,  based  on  reports  from  its  Member  States  (66).  This  is 
surprisingly close to the WHO estimate of 160,946 for drug use disorder deaths in 2013, but less than 
half of WHO draft estimates for total deaths from all causes attributable to drug use. 

8.15 Road injuries


For the third WHO Global status report on road safety  (67),  updated estimates of road injury deaths 
were  prepared  for  182  Member  States  for  the  years  2000‐2013.  These  estimates  drew  on  death 
registration data, on reported road traffic deaths from official road traffic surveillance systems (collected 
in  a  WHO  survey  of  Member  States  for  the  report),  and  on  a  revised  regression  model  for  countries 
without  useable  death  registration  data.  Road  injury  deaths  were  projected  forward  to  2015  using 
recent trends in death registration data where available, or the trend for recent years to 2015 from the 
GBD2015. Road injury deaths for Libya reported in the  third WHO Global status report on road safety  
were considerably higher than for any other country (based on country‐reported surveillance data) and 
were  revised  downwards  using  the  road  injury  regression  model  estimates  predicted  for  Libya  from 
relevant covariates such as vehicle ownership. 

8.16 Homicide
Updated estimates of homicide deaths for WHO Member States were published by WHO for years 2000‐
2012  in  the  Global  status  report  on  violence  prevention  2014  (68),  drawing  on  data  from  vital 
registration and criminal justice systems. These were projected forward to 2015 using recent trends in 
death registration data where available, or the trend for recent years to 2015 from the GBD2015. 

8.17 Conflict and natural disasters


Estimated  deaths  for  major  natural  disasters  were  obtained  from  the  EM‐DAT/CRED  International 
Disaster Database (69). These data were used to adjust age‐sex specific total as described in the WHO 
Life Tables Technical Paper (11).  
Country‐specific  estimates  of  war  and  conflict  deaths  were  updated  for  the  entire  period  1990‐2015 
using  revised  methods  documented  previously  which  draw  on  information  on  conflict  intensity,  time 
trends,  and  mortality  obtained  from  a  number  of  war  mortality  databases.  The  revised  WHO  country‐
specific estimates of war and conflict deaths for the period 1990‐2015 make use of estimates of direct 
deaths from three datasets: Battle‐Related Deaths (version 5), Non‐State Conflict Dataset (UCDP version 
5),  and  One‐sided  Violence  Dataset  (UCDP  version  5)  from  1989  to  2014  (70‐72).  Using  these  three 
datasets,  instead  of  focusing  solely  on  battle‐related  deaths,  reduces  the  likelihood  that  overall  direct 
conflict deaths are underestimated.  
However,  it  is    likely  that  a  degree  of  undercounting  still  occurs  in  the  count‐based  datasets,  and  a 
revised  adjustment  factor  of  1.91  is  applied  to  the  annual  battle  death  main  estimates  for  state‐state 

World Health Organization Page 38


 
conflicts.  No  adjustments  were  applied  to  estimated  conflict  deaths  (main  estimates)  for  non‐state 
conflict  deaths,  and  one‐sided  violence.  Note  that  the  application  of  a  single  adjustment  factor  for  all 
state‐state conflicts may result in deaths for specific conflicts being over‐ or under‐estimated.  
 For several conflicts where more specific sources of information are available, these were used to revise 
estimated deaths: 
Iraq  Latest counts of reported deaths  in Iraq by the Iraq  Body Count (73) were  compared 
with  conflict  deaths  for  the  period  2003‐2006  estimated  from  the  Iraq  Family  Health 
Survey  2006  (74).  This  nationally  representative  survey  of  9,345  households  included 
questions  on  deaths  of  adult  siblings  of  respondents,  and  deaths  in  the  household. 
Sibling  deaths  were  used  to  estimate  adult  mortality  rates  using  the  Gakidou‐King 
method  (75).  Calendar  year  adjustment  factors  for  under‐reporting  in  the  Iraq  Body 
Count  data  ranged  from  3.3  (2003)  and  3.4  (2004)  to  2.3  (2006)  and  2.2  (2007).  An 
average adjustment factor of 2.17 was applied to Iraq Body Count data for more recent 
years to derive a time series of estimated total conflict deaths in Iraq. 
Syria   For Syria, excess mortality in 2011 and 2012 due to the conflict was taken into account 
based  on  UN  estimates  of  overall  conflict  deaths  by  month  and  age  distribution  of 
deaths (76, 77), as well as estimates by various human rights organizations (78, 79). 
West  Bank  and  Gaza  Strip.    Estimates  of  Israeli  and  Palestinian  deaths  were  derived  from  statistics 
published  by  the  Office  for  the  Coordination  of  Humanitarian  Affairs  (OCHA)  ‐ 
Occupied  Palestinian  Territory  (OPT)  (80)  and  the  Israeli  Center  for  Human  Rights  in 
the Occupied Territories (81). 
Deaths  due  to  landmines  and  unexploded  ordinance,  terrorist  events  and  legal  execution  were 
estimated as described in the 2016 Technical Paper on WHO life table methods ().Deaths from terrorist 
events were separately estimated for many countries without ongoing general conflict using data from 
the  Global  Terrorism  Database  (82).  Terrorism  deaths  from  this  database  were  not  added  to  conflict 
deaths  for  Iraq,  Pakistan,  Afghanistan  and  a  number  of  African  countries  to  avoid  potential  double 
counting. 
Legal  execution  deaths  are  included  in  this  cause  category  for  GHE2015.  Estimated  execution  deaths 
were  added  for  the  main  countries  using  capital  punishment  regularly  (China,  Iran,  Iraq,  DPR  Korea, 
Saudi  Arabia,  USA  and  Yemen),  from  UN  Human  Rights  Reports,  with  additional  information  from 
Amnesty International reports, Human Rights Watch reports and Wikipedia. 
 

World Health Organization Page 39


 
9 Other causes of death for countries without useable data
9.1 Cause of death estimates from the GBD2015 study
The  IHME developed covariate based estimation models for a large number of single causes as inputs to 
its overall estimation of numbers of deaths by country, cause, age and sex for years 1990‐2010 in the 
GBD2010 study  (83). Results  from  these models  were used  as  inputs to  WHO Global  Health  Estimates 
2000‐2012  for  causes  of  death  not  addressed  by  WHO  and  UN  Interagency  estimation  processes  and 
where death registration data did not provide sufficient detail or did not meet quality criteria for direct 
use for estimating deaths by cause.  
For detailed cause‐of‐death subcategories not directly derivable from the death registration data in the 
WHO Mortality Database, we have drawn on updated IHME analyses from the GBD2015 study (4‐6), as 
described below. For  example, deaths  in the  full ICD cause  list categories  and the  short list  categories 
were  used  to  impute  more  detailed  cause  distributions  for  some  GHE  cause  categories  (for  example, 
cerebrovascular disease type) using relevant cause fractions from the GBD2015 analyses. Note that the 
IHME modelling strategies do make use of the available death registration data as well as other sources 
of information on deaths, covariate regression modelling and also draw on patterns of causes of death 
for similar countries.  
A number of modelling strategies were used by IHME for causes of death depending on the availability 
of data and the epidemiology of the disease (5). For most major causes of death except HIV/AIDS and 
measles,  IHME  used  ensemble  modelling  to  create  a  weighted  average  of  many  individual  covariate‐
based models (ranging from hundreds to thousands in some cases) for each specific cause. IHME cause 
of  death  estimation  methods  are  thus  complex  and  highly  computer‐intensive.  The  overall  out‐of‐
sample predictive validity of the ensemble is usually not much different to that of the top‐ranked model, 
but uncertainty ranges are generally much wider and more plausible than for single models. 
To ensure that the results of all the single‐cause models summed to the all‐cause mortality estimate for 
each  age‐sex‐country‐year  group,  IHME  applied  a  final  step  called  CoDCorrect  to  rescale  the  cause‐
specific estimates (5). This was done using repeated random draws from the uncertainty distributions of 
each  single  cause  and  from  the  all‐cause  envelope,  and  proportionately  rescaling  each  single  cause 
estimate  so  they  collectively  summed  to  the  envelope  estimate.  The  overall  effect  is  to  “squeeze”  or 
“expand” causes with wider uncertainty ranges more than those with narrower uncertainty ranges. 
The overall process of preparing the “prior” set of estimates for all countries for years 2000‐2015 for the 
complete GHE cause list ensuring that inputs from WHO/UN sources and GBD2015 were consistent with 
the WHO all‐cause envelopes is summarized in Figure 9.1. These “prior” estimates were used “as is” for 
causes  of  death  at  ages  5  and  over  for  countries  without  death  registration  data  meeting  inclusion 
criteria, and also provided inputs to the preparation of GHE2015 estimates for India, under 5 deaths and 
inputs  for  specific  detailed  cause  breakdowns  for  certain  cause  groups  for  countries  with  death 
registration data. 
 
   

World Health Organization Page 40


 
Figure 9.1 Overview of the processes involved in the preparation of the GHE2015 “prior” estimates 
for all countries. Refer also to Figure 1.1 for further steps involved in the inclusion of this dataset in 
the final GHE2015 estimates. 

 
 

9.2 GHECorrect process


GBD2015 results, post‐CoDCorrect, were used as inputs to estimate cause fractions by country, age, sex 
and year for causes of death at ages five years and above for which death registration data and/or WHO 
and UN Interagency analyses (described in Sections 4 to 8) were not available. IHME results for priority 
causes  such  as  HIV,  TB,  malaria,  cancers,  maternal  mortality,  child  mortality  differ  to  varying  degrees 
from those of WHO and UN agency partners. In part, this reflects differences in modelling strategies, but 
also  the  inclusion  by  IHME  of  data  from  verbal  autopsy  (VA)  studies  which  has  been  mapped  to  ICD 
categories  using  IHME‐developed  computer  algorithms.  As  was  done  for  GBD2015,  we  carried  out  a 
“GHECorrect  process  to  ensure  that  cause  fractions  across  all  causes  added  to  1    by  age,  sex,  country 
and year, meaning that estimated numbers of deaths added across causes to the estimated total deaths 
by age, sex, country and year.  
Since  WHO  and  IHME  all‐cause  envelopes  (death  rates,  life  tables)  differed  significantly  for  some  age 
groups and countries,  the first  step was to  rescale estimated  deaths for all  cause categories  excluding 
disasters and conflict (referred to as mortality shocks) in the GBD2015 to match the estimated WHO all‐
cause deaths excluding shocks: 
 

World Health Organization Page 41


 
 

 
where  G ysac is the corrected number of GBD2015 deaths for location l, year y, sex s, age a, cause c, 
G ysac is the corresponding uncorrected number of GBD2015 deaths, and E ysa is the estimated GHE total 
deaths excluding shocks for location l, year y, sex s, and age a.
Causes were divided into three groups:  
W  WHO/interagency  causes  with  estimates  (tuberculosis,  HIV,  malaria,  rabies,  maternal 
causes, cancers, road injuries and homicide) 
S  mortality shocks (natural disasters and conflicts) 
R  other causes 
The adjustment factor required for corrected GBD estimates for group R is: 
 

 
∈ ∈

The  adjustment  factor  α  was  less  than  0.5  for  3.4%  of  location‐year‐sex‐age‐specific  estimates  and 
between  0.5  and  0.75  for  another  5.7%.    To  reduce  the  need  for  substantial  squeezing  of  GBD2015 
inputs, adjustments were made to the age distribution of deaths for causes in group W’ (apart from HIV 
and  maternal  causes)  as  follows.  Where  α ysa was  <0.5,  deaths  D ysac    for  causes  c  in  cause  group  W’ 
were  adjusted  downwards  by  the  factor  0.5*( D ysaW’    +  CG ysaW’  )/  ( D ysaW’    )  and  the  excess  deaths 
redistributed pro‐rata across other ages for that location l, year y, and sex s.  After these adjustments to 
age distributions of causes in W’, the adjustment factor  α was less than 0.5 for 1.7% of location‐year‐
sex‐age‐specific estimates and between 0.5 and 0.75 for another 3.3%.  
GBD2015‐derived estimates for causes in group R were squeezed up to 25% using an alpha value  
′ max ,  
 
where  αmin =0.75. Where further squeezing was required to match the WHO envelope, estimates were 
squeezed  in  both  groups  R  and  W’,  with  a  differential  squeezing  factor  δ  =  1.5  (causes  in  group  R 
squeezed by factor δxβ and those in group W’ by factor β) where β was calculated as: 

β ′ δ ′  
∈ ∈ ∈ ∈
 
Some additional adjustments were carried out for age distributions of group W’ causes for Zimbabwe, 
due to large differences between GBD2015 and GHE non‐HIV envelopes. For Zimbabwe,   αmin was set 
to 0.9 and δ to 1. The squeeze factor β was capped at a maximum value of 0.5/ δ (0.5 for Zimbabwe, 
0.333 for other countries). There were 10 countries where β initially exceeded 0.5/ δ for some age‐sex 
categories,  seven  of  these  were  high  HIV  countries,  and  the  others  were  small  islands.  This  mostly 
occurred at younger adult ages and for years before 2010. In these cases, a further pro‐rata squeeze was 
applied to groups W (including and maternal now) and R to match the WHO envelope for 161 country‐

World Health Organization Page 42


 
year‐sex‐age  categories  (an  average  of  1  age‐sex  category  in  each  year  for  each  of  10  countries).    For 
these cases, the average additional adjustment was 16% downwards. 

9.3 Other adjustments for specific causes in certain countries


As  described  in  a  previous  technical  paper,  HIV  mortality  rates  and  non‐HIV  mortality  rates  were 
explicitly estimated for 43 high HIV prevalence countries (11). The resulting non‐HIV mortality envelopes 
differed substantially from those in GBD2015 for 4 countries, resulting in Group I (infectious, maternal, 
neonatal and nutritional causes) and Group II (non‐communicable diseases) fractions and age patterns 
substantially  different  in  GHE2015  from  those  of  neighboring  countries  with  similar  non‐HIV  mortality 
levels.  For this reason, the GBD2015 Group I and Group II inputs were further adjusted for Côte d’Ivoire, 
Nigeria,  Sierra  Leone  and  Zimbabwe,  based  on  group  fractions  for  other  countries  in  the  region  with 
similar non‐HIV mortality levels..   
The numbers of liver cancer deaths and liver cirrhosis deaths attributable to hepatitis B, hepatitis C and 
alcohol derived from the GBD2015 corrected estimates were further revised as described in Section 8.6. 
Estimates of deaths attributable to drug use disorders derived from GBD2015 corrected estimates were 
further  revised  to  account  for  opioid‐related  deaths  included  in  the  “other  and  multiple  drug  use” 
category as described in Section 8.14. Various other adjustments were also made for specific causes as 
described in Section 8. 
 Based on the GBD2015 inputs, there were also a number of extreme outliers for specific causes in some 
countries. These were adjusted as follows: 

 Deaths  due  to  meningitis  and  encephalitis  in  Nepal  and  Bhutan  were  revised  downwards  by  a 
factor  based  on  the  ratio  of  GBD2015  estimated  death  rate  to  that  for  India  applied  to  the 
GHE2015 revised meningitis and encephalitis deaths derived from the Indian SRS data. 
 Chronic respiratory disease deaths for Papua New Guinea (the highest in the world in GBD2015) 
were revised downwards to a level 20% above the average for Vanuatu, Micronesia and Fiji. 
 Accidental poisoning deaths in Somalia were revised downward by 60% based on the difference 
between GBD2013 and GBD2015 estimates. 
 Skin disease deaths were revised downwards in Bahrein based on the average of the death rates 
for United Arab Emirates, Kuwait, Oman, Qatar and Saudi Arabia. 
 Endocrine  disease  death  rates  were  adjusted  downwards  in  Tajikistan  using  GBD2013‐based 
rates, which are similar to those for Afghanistan and Pakistan in GBD2015. 
 GBD2015 death rates for chronic respiratory diseases in Montenegro were extremely low, unlike 
those  in  the  surrounding  countries  (Serbia,  Bosnia  and  Herzegovina,  Croatia,  Albania,  and  the 
Former Yugoslav Republic of Macedonia). In contrast, both smoking rates and air pollution levels 
were  similar  for  Montenegro  and  these  countries.  The  CRD  death  rates  for  Montenegro  were 
adjusted to match those of Serbia. 

9.4 Comparisons of major cause groups and differences


Figure  9.2  provides  a  comparison  of  major  cause  group  death  rates  for  the  GBD2015  and  WHO  GHE 
results  for  year  2015  for  seven  broad  regional  groupings,  after  the  GHECorrect  process  and  the  other 
cause‐ and country‐specific adjustments. 
 

World Health Organization Page 43


 
Figure 9.2  Comparison of GHE and IHME death rates per 100,000 population, major cause groups, 2015. 

World Health Organization Page 44


 
10 Uncertainty of estimates
Many of the inputs to the GHE2015 estimates have explicit uncertainty ranges. However, there are some 
exceptions: WHO life tables are based on the UN Population Division’s World Population Prospects 2015 
life tables and these do not yet have explicit uncertainty ranges estimated. Additionally, there are some 
specific  cause  inputs  from  WHO  and  UN  sources  which  do  not  yet  estimate  quantitative  uncertainty 
ranges. Given the challenges associated with calculating coherent quantitative uncertainty intervals with 
the  available  input  data,  guidance  to  users  on  the  quality  of  underlying  death  registration  data  is 
available together with country estimates, using methods described below (section 10.1).  In addition, 
quantitative uncertainty ranges are available as part of the comprehensive GHE2015 estimates dataset 
on the WHO website. Methods for these uncertainty ranges, as well as an overview of the quality of the 
uncertainty analysis, are described below (section 10.2). 

10.1 Guidance on underlying data quality


General guidance on the level of evidence available for death estimates for years 2000‐2015 is based on 
the quality of death registration data available in the WHO Mortality Database. Countries are classified 
into  five  levels,  with  descending  quality  of  death  registration  data,  as  described  in  Table  10.1. 
Classification is based on three characteristics: 1) whether the data are reported by ICD code or with a 
summarized cause list, 2) the number of years of data available in the WHO mortality database, and 3) 
the  average  usability  of  the  available  data  in  the  period  2000‐present.  Usability  is  calculated  as  the 
product of the proportion of deaths assigned to a set of ill‐defined cause of death codes (Table 4.2) and 
100  less  the  percentage  completeness.  Because  it  is  not  possible  to  obtain  the  full  number  of  deaths 
assigned  to  ill‐defined  causes  of  death  when  countries  report  death  registration  data  using  a 
summarized cause list, a more stringent set of usability cutoffs were defined for these countries. 
 
Table 10.1 Criteria for classification of countries by quality of death registration data  
Category  Countries reporting data by ICD code Countries reporting data with a summarized 
cause list 
Green  Average usability ≥ 80% for available years 2000 + N/A 
(highest)  At least 5 years’ data available from 2005 
Light  Average usability ≥ 60% for available years 2000 + Average usability ≥ 80% for available years 2000 +
yellow  At least 5 years’ data available from 2005  At least 5 years’ data available from 2005 
Dark  Average usability ≥ 40% for available years 2000 + Average usability ≥ 60% for available years 2000 +
yellow     
Light  No/unusable death registration data
pink  Low HIV country 
Dark pink  No/unusable death registration
(lowest)  High HIV country 
 
   

World Health Organization Page 45


 
The following guidance to users is provided together with the country data download: 
Multiple years of national death registration data with high completeness and quality of 
cause‐of‐death assignment are  available.  Estimates  for  these  countries  may  be 
compared and time series may be used for priority setting and policy evaluation.  
Multiple  years  of  death  registration  data  are  available.    Data  have  low  completeness 
and/or  issues  with  cause‐of‐death  assignment  which  likely  affect  estimated  deaths  by 
cause and time trends. Estimates may be used for priority setting.  Use   estimates  for 
programme  evaluation  with  caution,  as  improvements  in  the  vital  registration  system 
may  affect  the  estimated  trends  in  cause‐specific  mortality.  Comparisons  among 
countries  should  be  interpreted  with  caution.  Light  yellow  denotes  moderate  quality 
issues and dark yellow denotes severe quality issues. 
Death registration data are unavailable or unusable due to quality issues. Estimates of 
mortality  by  cause  should  be  interpreted  with  caution.    Estimates  may  be  used  for 
priority setting,  however,  they are  not  likely to  be  informative for  policy  evaluation  or 
comparisons among countries. Dark pink denotes countries with high HIV prevalence. 
 

10.2 Uncertainty ranges


Additionally,  quantitative  uncertainty  ranges  have  been  released  for  the  following  WHO/UN  cause‐
specific estimates: 
 UN‐IGME estimates of neonatal, infant and child mortality (10) 
 WHO‐MCEE estimates of neonatal and child causes of death (84) 
 Tuberculosis (32, 85) 
 Malaria (39, 86) 
 HIV (UNAIDS estimates) (33) 
 Maternal deaths (16) 
 Road injury (67) 
 Homicide (68) 
In addition, the IARC Globocan database provides information on data sources and quality of inputs for 
seven categories of incidence data and six categories of mortality data, as well as six estimation methods 
for mortality (87). 
Uncertainty intervals for global health estimates frequently underestimate total uncertainty.  Calculated 
uncertainty  ranges  depend  on  the  assumptions  and  methods  used.  These  methods  typically  reflect 
some,  but  not  all,  sources  of  uncertainty  (7).  Most  methods  for  estimation  of  uncertainty  rely  on 
statistical techniques  to  assess  variations  across  observations and/or  take  into  account  sampling  error 
but  are  less  successful  in  dealing  with  unknown  systematic  bias  in  observations,  such  as  systematic 
misspecification  of  cause  of  death.  Uncertainty  around  model  specification  and  data  pre‐processing 
steps are also frequently omitted when uncertainty intervals are quantified.  
The  GBD2015  estimates  of  deaths  by  cause,  age,  sex,  year  and  country  also  include  estimates  of  95% 
uncertainty  ranges  that  take  account  of  some,  but  not  all  sources  of  uncertainty.  Their  uncertainty 
estimates include uncertainties that arise from sample sizes of data, adjustments to sources of all‐cause 

World Health Organization Page 46


 
mortality, all‐cause and cause‐specific model specifications, and the CoDCorrect process. The GBD2015 
uncertainty ranges do not include uncertainty in garbage code redistribution algorithms, uncertainty in 
covariates  used  in  cause  of  death  models,  uncertainty  in  the  attribution  of  underlying  cause  of  death 
through  the  death  registration  and  coding  process,  or  uncertainty  in  the  attribution  of  cause  of  death 
from verbal autopsy data.  
The most common way of understanding how well uncertainty intervals perform is to withhold data, fit 
the model, and compare the withheld data to the uncertainty interval for the same country, cause and 
year  (usually  called  cross‐validation).  This  method  may  overstate  the  performance  of  uncertainty 
intervals,  since  it  weights  data‐rich  regions  most  heavily  and  does  not  take  into  account  data  pre‐
processing.    Another  method  is  to  calculate  how  often  the  results  of  a  methodologically  sound 
estimation  method  fits  within  the  confidence  interval  of  another  methodologically  sound  estimation 
method.  To  this  end,  we  compared  GBD2015  results  to  the  uncertainty  intervals  calculated  for 
GBD2013, by country, sex, and cause for a shortlist of causes (Table 4.6).  We found that 65% of total 
deaths  by  country,  cause  and  sex  for  the  year  2013  from  GBD2015  fell  within  the  95%  intervals 
presented with GBD2013 (Table 10.1).  
This  indicates  that  the  GBD2013  confidence  intervals  performed  quite  poorly,  given  that  the  two 
estimation exercises share the same methodological framework.  As expected, the GBD2013 confidence 
intervals  were  most  likely  to  contain  the  GBD2015  estimate  for  the  data‐rich  regions.  The  GBD2013 
uncertainty  intervals  performance  also  varied  by  cause  of  death,  with  only  28%  of  GBD2015  total 
malaria  deaths  by  country  and  sex  falling  within  the  GBD2013  uncertainty  interval,  likely  due  to 
methodological revisions between the two rounds of GBD. 
 
Table 10.1  Percentage of GBD2015 country‐sex results for 2013 falling within GBD2013  
                     uncertainty intervals, selected causes available in both rounds, by indicator. 

Number of deaths by cause  65% 

Rate by cause  64% 

Percentage of deaths by cause  58% 

 
Nevertheless,  the  GBD  enterprise  is  the  only  source  of  comprehensive  uncertainty  estimates  for 
mortality by cause. These intervals may be interpreted as a minimum uncertainty level. As a source of 
guidance on uncertainty in the GHE, we  summarize the broad variations of uncertainty in the GBD2015 
estimates across cause categories and countries, with the latter grouped by data sources and methods. 
The  following  tables  summarize  relative  uncertainty  intervals  by  cause,  country  size  and  data  class. 
Separate upper and lower uncertainty intervals are given as per cent values. Thus for example upper and 
lower uncertainty estimates  of 12% and 14% for an estimate of 1000 deaths would correspond to a 95% 
uncertainty range of (880,1140). 
 
   

World Health Organization Page 47


 
Table 10.2 Average relative uncertainty (%) for all‐cause all‐ages mortality, by country size and data class 
      VR 90+**      VR  80‐90** China/India Other   High HIV
Size*  Lower  Upper    Lower  Upper Lower Upper Lower Upper    Lower  Upper
Small  4.5  4.7    7.8  8.3 11.9 13.9    27.4  42.0
Medium  1.7  1.8    4.4  4.7 11.9 13.9    20.8  28.6
Large  0.9  0.9    3.3  3.4 4.0 4.1 11.9 13.9    15.7  19.5

* Size categories based on average deaths per year 2000‐2015 tertiles: small <30,000, large>150,000 deaths/year 
** VR = Death registration data, completeness 80% to <90% and 90% or more 
 
Table 10.2 summarizes average GBD2015 estimated uncertainty ranges for all‐cause mortality estimates 
(total  for  all  ages).  Table  10.3  summarizes  relative  average  for  all‐age  estimates  of  cause‐specific 
mortality  for  four  categories  of  causes.  GBD2015  relative  uncertainty  ranges  correlate  quite  strongly 
with the magnitude of the number of deaths estimate, indicating the size of the uncertainty interval is 
driven  stochastic  error  rather  than    systematic  biases  which  would  be  less  correlated  with  the  total 
number  of  deaths.  Thus  Table  10.3  presents  summary  uncertainty  estimates  for  a  selection  of 
magnitudes ranging from 10 deaths per year to 1 million deaths per year. 
Uncertainty  ranges  for  age‐specific  mortality  estimates  are  wider  than  those  for  all‐cause  mortality, 
except  for  older  age  groups  in  countries  with  useable  death  registration  data.  Relative  uncertainty 
multipliers  for  specific  age  groups  are  given  in  Table  10.4  by  cause  category  and  data  class.  These 
multipliers are applied to the all‐age uncertainty ranges given in Tables 10.2 and 10.3 to produce age‐
specific uncertainty estimates. 
 
 

World Health Organization Page 48


 
Table 10.3 Average relative uncertainty (%) for all‐ages mortality, by cause category, magnitude of deaths and data class 
VR  90+* VR  80‐90* Other High HIV
Cause  Deaths per year Lower Upper Lower  Upper Lower Upper Lower Upper
Uncertainty category 1  10 18 19 18  20 25 32 72 136
Includes 1st level cause groups I, II, III as well as 2nd  100 11 12 13  14 21 27 52 89
level cause groups totals for malignant neoplasms,  1,000 7 7 9  10 18 22 38 58
mental disorders and substance use, cardiovascular 
10,000 4 4 7  7 16 18 28 38
diseases 
100,000 3 3 5  5 14 15 20 25
1,000,000 2 2 3  3 12 13 15 16
Uncertainty category 2. Includes 2nd level causes  10 16 18 18  22 29 38 50 83
neonatal, neurological, chronic respiratory, digestive,  100 13 14 15  17 26 33 42 64
musculoskeletal, unintentional injuries, intentional  1,000 10 10 13  14 23 28 36 50
injuries, and 3rd level causes in the unintentional 
10,000 8 8 11  11 21 24 30 38
injuries (except poisoning), cancers, neurological 
disorders (except dementia), cardiovascular conditions  100,000 6 6 9  9 19 21 25 30
(except hyper‐tensive heart disease), and COPD,  1,000,000 5 5 7  7 17 18 22 23
alcohol and drug use disorder (total)   
Uncertainty category 3.  Includes 2nd level causes   10 21 25 24  31 34 48 51 92
meingitis, encephalitis, hepatitis,parasitic/vector  100 17 19 20  24 31 43 46 75
diseases, other infectious diseases, endocrine, blood  1,000 14 15 18  19 29 38 41 62
and immune disorders, congenital anomalies and 3rd 
10,000 12 11 15  15 27 34 37 50
level causes in the digestive, genitourinary, 
musculoskeletal and intentional injury groups as well  100,000 10 9 13  12 26 30 33 41
as eating disorders, dementia,hypertensive heart  1,000,000 8 7 11  9 24 27 30 34
disease, asthma and specific drug use disorders   
Uncertainty category 4.   Includes 3rd level causes in  10 32 43 41  61 53 97 65 137
the following 2nd level cause groups: STDs, childhood‐ 100 25 31 37  51 51 89 61 120
cluster, hepatitis, parasitic/vector, intestinal  1,000 20 22 34  42 50 83 57 105
nematode, neonatal conditions, nutritional 
10,000 16 15 32  35 49 77 54 92
deficiencies, endocrine, blood and immune disorders, 
congenital anomalies, as well as leprosy, upper  100,000 13 11 29  29 48 71 50 80
respiratory tract infections, SIDS and poisoning  1,000,000 10 8 27  24 47 65 47 70

* VR = Death registration data, completeness 80% to <90% and 90% or more 

World Health Organization Page 49


 
Table 10.4 Relative uncertainty multipliers of all‐ages uncertainty for specific age groups, by cause category, and data class  

Age group  VR  90+** VR  80‐90** Other High HIV


Cause categories  (years)*  Lower Upper Lower  Upper Lower Upper Lower Upper
All causes   5‐14  3.32  3.61    2.56  2.96    1.19  1.24    1.05  1.10 
  15‐29  1.56  1.63    1.49  1.67    1.47  1.51    1.93  1.86 
  30‐49  1.40  1.44    1.31  1.35    1.59  1.69    1.75  1.85 
  50‐59  1.23  1.25    1.16  1.15    1.51  1.56    2.00  2.28 
  60‐69  1.15  1.15    1.06  1.03    1.34  1.34    1.90  2.02 
  70+  0.89  0.88    0.82  0.80    1.00  1.00    1.42  1.32 
Uncertainty category 1   5‐14  3.99  4.31    3.20  3.70    1.80  1.92    1.03  1.05 
Includes 1st level cause groups I, II, III as well as 2nd 
level cause groups totals for malignant neoplasms,  15‐29  2.97  3.19    2.43  2.71    1.74  1.91    1.49  1.60 
mental disorders and substance use, cardiovascular  30‐49  2.23  2.36    2.01  2.13    1.67  1.83    1.46  1.68 
diseases 
50‐59  1.88  2.00    1.75  1.84    1.52  1.65    1.44  1.64 
60‐69  1.72  1.82    1.61  1.68    1.40  1.48    1.33  1.40 
70+  1.68  1.75    1.51  1.58    1.26  1.29    1.12  1.08 
Uncertainty category 2. Includes 2nd level causes  Neonatal 2.36  2.51    2.58  3.03    1.91  2.59    1.49  1.99 
neonatal, neurological, chronic respiratory, digestive,  1‐59 months
musculoskeletal, unintentional injuries, intentional  3.43  4.15    3.36  4.43    2.41  3.52    1.74  2.43 
injuries, and 3rd level causes in the unintentional  5‐14  2.62  3.05    2.58  3.14    1.74  2.06    1.22  1.29 
injuries (except poisoning), cancers, neurological 
15‐29  2.55  2.96    2.41  2.88    1.73  2.06    1.40  1.63 
disorders (except dementia), cardiovascular conditions 
(except hyper‐tensive heart disease), and COPD,  30‐49  2.28  2.59    2.17  2.47    1.71  1.98    1.39  1.71 
alcohol and drug use disorder (total)  50‐59  2.10  2.35    2.03  2.27    1.58  1.81    1.34  1.59 
60‐69  2.00  2.22    1.96  2.18    1.51  1.69    1.26  1.38 
70+  1.97  2.17    1.92  2.16    1.47  1.63    1.16  1.20 
(continued on next page) 
   

World Health Organization Page 50


 
Table 10.4 (continued) Relative uncertainty multipliers of all‐ages uncertainty for specific age groups, by cause category, and data class 

Age group  VR  90+** VR  80‐90** Other High HIV


Cause categories  (years)  Lower Upper Lower  Upper Lower Upper Lower Upper
Uncertainty category 3.  Includes 2nd level causes   Neonatal 2.11  2.32    1.99  2.32    1.70  2.10    1.44  1.71 
meningitis, encephalitis, hepatitis, parasitic/vector  1‐59 months
diseases, other infectious diseases, endocrine, blood  2.45  2.93    2.43  3.06    1.99  2.75    1.57  2.10 
and immune disorders, congenital anomalies and 3rd  5‐14  2.32  2.74    2.22  2.64    1.58  1.79    1.21  1.25 
level causes in the digestive, genitourinary, 
15‐29  2.11  2.54    2.07  2.47    1.55  1.78    1.33  1.49 
musculoskeletal and intentional injury groups as well 
as eating disorders, dementia,hypertensive heart  30‐49  1.96  2.28    1.93  2.26    1.58  1.82    1.36  1.64 
disease, asthma and specific drug use disorders  50‐59  1.91  2.20    1.89  2.18    1.53  1.76    1.35  1.60 
60‐69  1.85  2.10    1.85  2.12    1.50  1.70    1.30  1.46 
70+ 1.79  2.01    1.82  2.11    1.47  1.66    1.22  1.31 
Uncertainty category 4.   Includes 3rd level causes in  Neonatal 1.43  1.53    1.27  1.40    1.26  1.43    1.20  1.28 
the following 2nd level cause groups: STDs, childhood‐ 1‐59 months
cluster, hepatitis, parasitic/vector, intestinal  1.65  1.89    1.56  1.93    1.42  1.97    1.32  1.79 
nematode, neonatal conditions, nutritional  5‐14  1.66  1.89    1.52  1.86    1.24  1.38    1.14  1.22 
deficiencies, endocrine, blood and immune disorders, 
15‐29  1.80  2.02    1.67  2.03    1.30  1.52    1.24  1.47 
congenital anomalies, as well as leprosy, upper 
respiratory tract infections, SIDS and poisoning  30‐49  1.73  1.89    1.58  1.91    1.34  1.63    1.28  1.60 
50‐59  1.71  1.90    1.65  1.99    1.32  1.61    1.28  1.58 
60‐69  1.64  1.83    1.62  1.94    1.32  1.60    1.27  1.53 
70+ 1.62  1.77    1.56  1.90    1.28  1.51    1.22  1.42 
* Uncertainty ranges for total neonatal and under 5 deaths are given elsewhere (10), as are uncertainty ranges for major causes of child death (84) 
** VR = Death registration data, completeness 80% to <90% and 90% or more 
 

World Health Organization Page 51


 
11 Conclusions
GHE2015 presents results for 183 WHO Member States, encompassing all those with a population of 90,000 
or greater in 2015. The GHE2015 estimates of causes of death by country, region and world for years 2000‐
2015 confirm and expand previous WHO analyses of global health trends and improvements.  In particular, 
WHO  published  an  assessment  of  progress  towards  and  achievement  of  the  UN  Millenium  Development 
Goals at the end of 2015 (88) and followed this by the World Health Statistics 2016 (89), which focused on 
progress and challenges for the UN Sustainable Development Goals and Targets for year 2030. 
The Sustainable Development Goals expand the focus of health targets from the unfinished agenda for child 
and  maternal  mortality,  and  infectious  diseases  to  a  broader  agenda  also  including  non‐communicable 
diseases,  injuries,  health  emergencies  and  health  risk  factors,  as  well  as  a  strong  focus  on  universal  health 
care (UHC)  (90,  91).  These  GHE2015 estimates  of  trends  and  levels  of mortality  by  cause  will  contribute  to 
WHO and UN monitoring and reporting of the health SDG goal and targets. 
WHO’s  adoption  of  health  estimates  is  affected  by  a  number  of  factors,  including  a  country  consultation 
process for country‐level health estimates, existing multi‐agency and expert group collaborative mechanisms, 
and compliance with standards around reporting data and methods.  More detailed information on quality of 
data sources and methods, as well as estimated uncertainty intervals, is provided in in this document and in 
other  referenced  sources.  As  required  by  the  GATHER  guidelines  (7),    documentation  of  data  inputs, 
methods, and results, including  uncertainty, has improved (Annex C provides the location of each GATHER 
reporting item). 
The  type  and  complexity  of  models  used  for  global  health  estimates  varies  widely  by  research/institutional 
group  and  health  estimate.  More  complex  models  are  necessary  to  generate  more  accurate  uncertainty 
intervals. As expected, these require greater researcher expertise and time and computational resources to 
run. Where  data  are  available  and  of  high  quality,  estimates  from  different  institutions  are  generally  in 
agreement.    Discrepancies  are  more  likely  to  arise  for  countries  where  data  are  poor  and  for  conditions 
where data are sparse and potentially biased. This is best addressed through improving the primary data.
Country  health  information  systems,  including  vital  registration,  need  to  be  strengthened  as  a  matter  of 
priority, in order to provide a more solid empirical basis for monitoring health situation and trends. Such data 
are also crucial for Member States’ monitoring of national and sub‐national trends in order to respond to the 
changing needs of their populations. 
To improve monitoring of mortality,  morbidity and risk factors health  information systems should focus  on 
strengthening: 
 Death  registration  through  civil  registration  and  vital  statistics  systems  (CRVS),  local  health  and 
demographic studies and other sources 
 Cause of death data collection through vital registration and verbal autopsy in communities 
 Regular household health surveys that include biological and clinical data collection 
 Complete facility recording and reporting with regular quality control 
 
 

11.1 Leading causes of death in 2015


Of  the  56.4  million  deaths  worldwide  in  2015,  more  than  half  (54%)  were  due  by  the  following  10  causes 
(Figure 11.1). Ischaemic heart disease and stroke killed 15 million people in 2015 – these two diseases have 
remained the biggest killers globally in the last 15 years.  

World Health Organization Page 52


 
Chronic  lung  disease  claimed  3.2  million  lives  in  2015,  while  lung  cancer  (along  with  trachea  and  bronchus 
cancers) caused 1.7 million deaths. Diabetes killed 1.6 million people in 2015, up from less than 1 million in 
2000. Deaths due to dementias more than doubled between 2000 and 2015, making it the 7th leading cause 
of global deaths in 2015.  
Lower  respiratory  infections  remained  the  most  deadly  communicable  disease,  causing  3.2  million  deaths 
worldwide in 2015. Diarrhoea death rate almost halved between 2000 and 2015, but the disease still caused 
a large number of deaths (1.4 million) in 2015. Similarly, tuberculosis death rate decreased  during the same 
period, but the disease was still among the top 10 causes of death in 2015 with a death toll of 1.4 million. 
Road injury killed 1.3 million people in 2015, three‐quarter of which were men and boys. 
 
Figure 11.1 The 10 leading causes of death in the world, 2015 

  0%   
 
   

World Health Organization Page 53


 
Figure 11.2 Leading causes of death by country income group, 2015 

 
 
 
   

World Health Organization Page 54


 
More than half (52%) of all deaths in low‐income countries in 2015 were caused by the so‐called “Group I” 
conditions,  which  include  communicable  diseases,  maternal  causes,  conditions  arising  during  the  perinatal 
period, and nutritional deficiencies. By contrast, less than 7% of deaths in high‐income countries were due to 
such causes. Lower respiratory infections were among the leading causes of death across all income groups 
(Figure 11.2). 
Noncommunicable diseases (NCDs) caused 70% of deaths globally, with regional figures ranging from 37% in 
low‐income  countries  to  88%  in  high‐income  countries.  Almost  all  of  the  10  leading  causes  of  death  in  the 
latter  group  of  countries  were  NCDs.  In  terms  of  absolute  number  of  deaths,  however,  78%  of  global  NCD 
deaths occurred in low‐ and middle‐income countries. 
Injuries claimed nearly 5 million lives in 2015. Over a quarter (27%) of these deaths were due to road traffic 
injuries. Low‐income countries had the highest mortality rate due to road traffic injuries with 28.5 deaths per 
100  000  population    –    the  global  rate  was  18.3.  Road  injuries  were  also  among  the  leading  10  causes  of 
death in both lower‐middle‐ and upper‐middle‐income countries. 
 

11.2 Reasons for changes in GHE estimates in this revision


As  with  previous  revisions  of  WHO  GHE  and  specific‐cause  time  series  estimates,  GHE2015  provides  an 
update  for  the  entire  time  series  from  2000  to  2015  incorporating  data  sources  and  specific 
WHO/interagency  and  IHME  estimates  released  since  GHE2013.  This  time  series  supercedes  previous  GHE 
time series, and differences between revision series should not be interpreted as time trends. 
Major  causes  of  significant  changes  in  estimates  or  trends  for  individual  countries  or  for  specific  causes  at 
country, regional or global level include the following: 
 The revision of WHO life table time series for Member States to align more closely with the UN WPP 
2015 life table time series (8). This resulted in substantial changes for some countries, and also for 
some  high  HIV  countries  due  to  significant  changes  in  UN  WPP  and  UNAIDS  modelling  of  the  HIV 
epidemics in these countries. 
 Significant  updates  to  time  series  estimates  for  WHO  and  UN  interagency  estimates  for  specific 
causes  such  as  HIV,  tuberculosis,  and  malaria,  as  well  as  incorporation  of  new  WHO  estimates  for 
causes such as homicide and road injury. 
 Revision of maternal mortality estimates to take account of revisions to all‐cause mortality envelopes 
in the reproductive age range 15‐49 years. 
 Use of more recent SRS data on causes of death in India, allowing the estimation of time trends from 
the two data sets for 2001‐2003 and 2010‐2013. 
 Use of the recently published IHME GBD 2015 study for GHE2015. There were substantial changes in 
GBD 2015 estimates for some causes and  countries compared to the previous GBD 2010  estimates 
used for GHE2013. 
 Improvements to the GHECorrect process (Section 9.2) used to ensure that cause‐specific estimates 
summed to WHO all‐cause mortality estimates derived from WHO life tables. 
 Substantial improvements in the availability of death registration data for many countries, together 
with improvements in the mapping of ICD‐based cause groups to the GHE cause groups. 

World Health Organization Page 55


 
 Substantial revision of the garbage code redistribution algorithms used for ill‐defined cardiovascular 
diseases and hypertensive heart disease, plus additional adjustments to deal with high proportions of 
deaths coded to “other” categories for certain ICD chapters in some countries. 
 Greater  reliance  on  estimates  derived  from  death  registration  data  to  supercede  older  WHO 
estimates for selected causes, particularly for cancers. 

11.3 Limitations of GHE estimates


Here  we  highlight  some  broad  cross‐cutting  limitations  to  the  GHE  mortality  and  cause  of  death  analysis. 
Comparable  information  about  death  numbers  and  rates  by  age,  sex,  cause,  year,  and  country  provides 
important  information  for  priority  setting  discussions  and  for  monitoring  and  evaluating  progress  towards 
global health goals. Major limitations and challenges are summarized below. 
 All‐cause mortality estimates in countries without well‐functioning death registration systems relies 
heavily on census and survey data sources (particularly sibling survival data) and the use of model life 
tables.  There  is  not  yet  consensus  on  the  methods  for  analyzing  sibling  survival  data  or  assessing 
levels of under‐reporting of deaths in surveys or censuses.  
 Demographic  methods  for  the  assessment  of  completeness  of  death  registration  all  involve  strong 
assumptions or information about migration and are prone to error resulting from age mis‐statement 
in registration or census data, and to differential completeness of successive censuses. 
 Estimation  of  HIV  mortality  relies  on  imputation  of  deaths  from  seroprevalence  data  using  limited 
information  on  survival  curves  for  HIV‐positive  persons  not  receiving  or  receiving  anti‐retroviral 
treatment  (ART),  and  on  the  coverage  of  ART  in  populations.  This  results  in  large  uncertainty  for 
countries with high prevalence of HIV, as disease progression rates may well vary across countries. 
 Although  death  registration  data  is  generally  the  best  form  of  information  available  on  causes  of 
death, it has considerable limitations, even in well‐functioning systems with medical certification of 
cause  of  death.  The  so‐called  garbage  codes  represent  a  substantial  proportion  of  deaths  in  some 
countries,  and  methods  for  re‐assigning  these  deaths  to  valid  causes  are  highly  uncertain  and 
generally are not based on empirical data. The assignment of underlying cause of death is limited by 
the information provided on the death certificate and quite sensitive to the order in which diagnoses 
are  written.  For  most  causes  of  death,  variability  (due  to  differences  in  physician  practice  when 
certifying a death) in assignment of valid causes of underlying death has not been addressed to date. 
Additionally,  some  diseases  and  injuries  have  specific  problems  associated  with  difficulty  in  making 
causal  judgments  of  underlying  cause  (eg.  diabetes  and  heart  disease,  or  Alzheimer’s  disease  and 
heart disease, drug or alcohol overdose). Finally, HIV and other stigmatized causes of death, such as 
suicide, are routinely miscoded; the miscoding rate varies by setting. 
 For  many  countries  without  functioning  death  registration  systems,  particularly  in  Africa,  there  is 
strong reliance on verbal autopsy studies, most of which are not nationally representative samples. 
Until  recently  there  has  been  considerable  variation  in  verbal  autopsy  instruments,  and  in  analysis 
and cause assignment methods. Validation studies are challenging, and difficult to generalize to other 
settings. The Indian SRS data included in the Million Death Study use a form of physician‐assignment 
of  underlying  cause  that  may  be  subject  to  different  biases  and  limitations  than  the  statistical 
algorithms used in InterVA or Tariff analyses. 
 The WHO GHE estimates bring together single cause analyses from a number of WHO departments, 
interagency  collaborations,  and  other  sources,  together  with  estimates  drawn  from  the  IHME 
GBD2015 study. These estimates are updated on differing time tables, and using different methods 

World Health Organization Page 56


 
and assumptions in some cases, and it is more difficult to ensure consistency across causes, than is 
the  case  for  large  comprehensive  estimates  such  as  GBD2015  prepared  by  a  single  study  group.  In 
addition, separate preparation of estimates of total mortality and cause‐specific mortality can lead to 
incompatible  cause‐specific  and  total  mortality  estimates.  In  some  cases,  WHO/UN  estimates  are 
prepared  only  for  all‐age  deaths,  and  age  patterns  imputed  from  available  sometimes  limited 
evidence. 
 Estimates  of  deaths  associated  with  mortality  shocks  (mainly  conflict  and  disasters,  but  also  some 
epidemics) are highly uncertain, and age patterns are generally imputed from limited data for other 
shocks.  Additionally,  in  countries  without  functioning  death  registration  systems  or  high  quality 
censuses,  it  is  very  difficult  to  take  account  of,  and  to  estimate,  indirect  mortality  associated  with 
mortality shocks, with increases in non‐injury mortality rates associated with disruption to health and 
other social systems.  
 While the uncertainty estimates discussed in Section 10 provide some guidance on the limitations of 
interpretation of the results, it should be kept in mind that these estimates reflect a subset of sources 
of uncertainty, and true uncertainty is higher. 
 

World Health Organization Page 57


 
References
(1)   Global Health Estimates: Deaths by cause, age and sex, by country and by region, 2000‐2012. Geneva: World 
Health Organization; 2014 (http://www.who.int/healthinfo/global_burden_disease/estimates_previous/en/ , 
accessed 16 December 2016). 
(2)   International Classification of Diseases – 10th Revision. Geneva: World Health Organization; 1990. 
(3)  Global Health Estimates 2015: Deaths by cause, age and sex, by country and by region, 2000‐2015. Geneva: 
World Health Organization; 2016.  (http://www.who.int/healthinfo/global_burden_disease/en/ accessed 16 
December 2016) 
(4)  GBD2015 Mortality and Causes of Death Collaborators. Global, regional, and national life expectancy, all‐cause 
mortality, and cause‐specific mortality for 249 causes of death, 1980–2015: a systematic analysis for the Global 
Burden of Disease Study 2015. Lancet 2016;388(10053):1459–1544. 
(5)   Supplementary Appendix. Supplement to: GBD2015 Mortality and Causes of Death Collaborators. Global, 
regional, and national life expectancy, all‐cause mortality, and cause‐specific mortality for 249 causes of death, 
1980–2015: a systematic analysis for the Global Burden of Disease Study 2015. Lancet 2016;388(10053):1459–
1544. 
(6)  Global Burden of Disease Study 2015. Global Burden of Disease Study 2015 (GBD2015) Results. Seattle, United 
States: Institute for Health Metrics and Evaluation (IHME), 2016. Datasets downloaded from the GBD Results Tool 
(http://ghdx.healthdata.org/gbd‐results‐tool, accessed 11 October 2016). 
(7)   The GATHER Working Group. Guidelines for Accurate and Transparent Health Estimates Reporting: the GATHER 
statement. Lancet 2016;388(10062):e19‐e23. 
(8)  World Population Prospects: the 2015 revision. New York (NY): United Nations, Department of Economic and 
Social Affairs, Population Division; 2015. 
(9)  WHO Mortality Database. Geneva: World Health Organization 
(http://www.who.int/healthinfo/mortality_data/en/ , accessed 31 October 2016). 
(10)  Levels & trends in child mortality: Report 2015. UNICEF, WHO, the World Bank, United Nations Population 
Division. New York: UNICEF on behalf of the UN Inter‐agency Group for Child Mortality Estimation; 2015 
(http://www.childmortality.org/files_v20/download/IGME%20report%202015%20child%20mortality%20final.pdf
 , accessed 9 April 2016). 
(11)   WHO methods and data sources for life tables 1990‐2015. Global Health Estimates Technical Paper 
WHO/HIS/IER/GHE/2016.2. Geneva: World Health Organization; 2016 
(http://www.who.int/healthinfo/statistics/LT_method.pdf?ua=1 , accessed 15 December 2016). 
(12)  Institute for Health Metrics and Evaluation. Causes of death (COD) visualization. 
(http://vizhub.healthdata.org/cod/ , accessed 15 December 2016). 
(13)   Mathers CD, Lopez AD, Murray CJL, Ezzati M, Jamison DT. The burden of disease and mortality by condition: data, 
methods and results for 2001. Global burden of disease and risk factors. New York, Oxford University Press, 2006. 
p. 45–240. 
(14)  Ahern RM, Lozano R, Naghavi M, Foreman K, Gakidou E, Murray CJ. Improving the public health utility of global 
cardiovascular mortality data: the rise of ischemic heart disease. Population Health Metrics 2011;9(1):1. 
(15)  Stevens GA, King G, Shibuya K. Deaths from heart failure: using coarsened exact matching to correct cause‐of‐
death statistics. Population Health Metrics 2010;8(1):1. 
(16)   Trends in maternal mortality: 1990 to 2015. Geneva: World Health Organization on behalf of WHO, UNICEF, 
UNFPA, The World Bank and the United Nations Population Division; 2015. 
(17)   Australian Bureau of Statistics. Causes of Death, Australia, 2015. Complexities in the measurement of Bowel 
Cancer in Australia, 2006. Available at:  

World Health Organization Page 58


 
http://www.abs.gov.au/ausstats/[email protected]/Lookup/by%20Subject/3303.0~2015~Main%20Features~Complexities
%20in%20the%20measurement%20of%20bowel%20cancer%20in%20Australia~7 
(18)  Liu L, Oza S, Hogan D, Chu Y, Perin J, Zhu J, Lawn JE, Cousens S, Mathers C, Black  RE. Global, regional and national 
causes  of  child  mortality  in  2000‐2015  –  an  updated  systematic  analysis  with  implications  for  the  Sustainable 
Development Goals. The Lancet, November 2016 doi: http://dx.doi.org/10.1016/S0140‐6736(16)31593‐8 
(19)  MCEE‐WHO methods and data sources for child causes of death 2000‐2015. Geneva: World Health Organization; 
2016 (http://www.who.int/healthinfo/global_burden_disease/ChildCOD_method_2000_2015.pdf?ua=1 , 
accessed 15 December 2016). 
(20)   Liu L, Oza S, Hogan D, Perin J, Rudan I, Lawn JE, Cousens S, Mathers C, Black R. Global, regional and national 
causes of child mortality in 2000‐2013 with projections to inform post‐2015 priorities: an updated systematic 
analysis. The Lancet 2015, Jan 31; 385(9966): 430‐440. 
(21)  Liu L, Johnson HL, Cousens S, Perin J, Scott S, Lawn JE, Rudan I, Campbell H, Cibulskis R, Li M, Mathers C, Black RE, 
for the Child Health Epidemiology Reference Group of WHO and UNICEF. Global, regional, and national causes of 
child mortality: an updated systematic analysis for 2010 with time trends since 2000. Lancet 2012;379:2151‐61. 
(22)  Bassani DG, Kumar R, Awasthi S et al. Causes of neonatal and child mortality in India: a nationally representative 
mortality survey. Lancet, 2010;376(9755):1853‐1860.  
(23)   China Ministry of Health‐Unpublished tabulations ‐‐ Vital Registration System cause‐of‐death data submitted 
annually to WHO. 
(24)   全国疾病监测系统死因监测数据集 [National Disease Surveillance System monitoring causes of death 2010]. 
Chinese Center for Disease Control and Prevention, Beijing, Military Medical Science Press, 2012, ISBN 978‐7‐
80245‐827‐7. 
(25)   中国疾病监测报告[Cause‐of‐death data from Chinese Disease Surveillance Points], China Ministry of Health, 
2004‐2009.  
(26)  China Centre for Disease Control and Prevention (CDC). Death Registration System ‐Unpublished tabulations for 
year 2013. 
(27)  Zhou M, Wang H, Zhu J, et al. Cause‐specific mortality for 240 causes in China during 1990–2013: a systematic 
subnational analysis for the Global Burden of Disease Study 2013. Lancet 2016; 387: 251–72. 
(28)   Registrar General of India. Causes of Death in India in 2001‐2003. New Delhi, Registrar General of India, 
Government of India, 2009. 
(29)   Jha P, Gajalakshmi V, Gupta PC, Kumar R, Mony P, Dhingra N et al. Prospective study of one million deaths in 
India: rationale, design, and validation results. PLoS Med, 2006;3(2):e18. 
(30)  Aleksandrowicz L, Malhotra V, Dikshit R, Gupta P, Kumar R, Sheth J, Rathi S, Suraweera W, Miasnikof P, Jotkar R, 
Sinha D, Awasthi S, Bhatia P, Jha P. Performance criteria for verbal autopsy‐based systems to estimate national 
causes of death: development and application to the Indian Million Death Study. BMC Medicine 2014, 12:21 
(31)  Registrar General of India and the Centre for Global Health Research. Causes of Death Statistics for India 2010‐
2013. A Joint Report of the Registrar General of India and the Centre for Global Health Research.  Government of 
India  Ministry  of  Home  Affairs  –  Sample  Registration  System,  2015.  (http://www.censusindia.gov.in/2011‐
common/Sample_Registration_System.html ) 
(32)  Global Tuberculosis Report 2016. Geneva, WHO, 2015. Geneva: World Health Organization; 2016. 
http://www.who.int/tb/publications/global_report/en/ 
(33)  HIV estimates with uncertainty bounds 1990‐2015. Geneva, UNAIDS; 2016.  
(http://www.unaids.org/en/resources/documents/2016/HIV_estimates_with_uncertainty_bounds_1990‐2015 , 
accessed 28 August 2016). 
(34)  Stover J, Andreev K, Slaymaker E, et al. Updates to the Spectrum model to estimate key HIV indicators for adults 
and children. AIDS (London, England). 2014;28(4):S427‐S434. doi:10.1097/QAD.0000000000000483. 

World Health Organization Page 59


 
(35).   Meek SR. Epidemiology of malaria in displaced Khmers on the Thai‐Kampuchean border. Southeast Asian J Trop 
Med Public Health, 1988;19(2):243‐52. 
(36)  X2. Luxemburger C, Ricci F, Nosten F, Raimond D, Bathet S, White NJ. The epidemiology of severe malaria in an 
area of low transmission in Thailand. Trans R Soc Trop Med Hyg, 1997;91(3):256‐62. Epub 1997/05/01. 
(37)  X3.  Alles  HK,  Mendis  KN,  Carter  R.  Malaria  mortality  rates  in  South  Asia  and  in  Africa:  implications  for  malaria 
control. Parasitol Today, 1998;14(9):369‐75. 
(38)  Douglas NM, Pontororing GJ, Lampah DA, Yeo TW, Kenangalem E, Poespoprodjo J, et al. Mortality attributable to 
Plasmodium vivax malaria: a clinical audit from Papua, Indonesia. BMC Med. 2014;12(1):217. 
(39)  World Malaria Report 2016. Geneva: World Health Organization; 2016. 
http://www.who.int/malaria/publications/world‐malaria‐report‐2016/en/ 
(53)  World Health Organization. World Malaria Report 2008. Geneva, World Health Organization, 2008. 
(40)  WHO and UNICEF estimates of national immunization coverage. 
( http://www.who.int/immunization_monitoring/routine/immunization_coverage/en/index4.htm). 
(41)  Final Report:  Global Pertussis Burden Expert Elicitation. 2011. Geneva, World Health Organization; 2011. 
(42)   Global Burden of Polio in Report on the WHO Quantitative Immunization and Vaccines Related Research (QUIVER) 
Advisory Committee Meeting Geneva, 4‐6 October 2011. WHO/IVB/12.03. Geneva, World Health Organization; 
2012. 
(43)   Chen  S,  Fricks  J,  Ferrari  MJ.  Tracking  measles  infection  through  non‐linear  state  space  models.  Journal  of  the 
Royal Statistical Society Series C, 2011, 61 (1). 
(44)  Simons E, Ferrari M, Fricks J, Wannemuehler K, Anand A, Burton A, et al. Assessment of the 2010 global measles 
mortality reduction goal: results from a model of surveillance data. Lancet, 2012, 379(9832):2173‐8. 
(45)  Wolfson LJ, Strebel PM, Gacic‐Dobo M, Hoekstra EJ, McFarland JW, Hersh BS. Has the 2005 measles mortality 
reduction goal been achieved? A natural history modelling study. Lancet, 2007,369(9557):191‐200. 
(46)  Sudfeld CR, Halsey NA. Measles Case Fatality Ratio in India: A Review of Community Based Studies. Indian 
Pediatrics, 2009, 46(11):983‐9. 
(47)   Joshi  AB,  Luman  ET,  Nandy  R,  Subedi  BK,  Liyanage  JBL,  Wierzba  TF.  Measles  deaths  in  Nepal:  estimating  the 
national case–fatality ratio. Bulletin of the World Health Organization, 2009, 87(6):456–465. 
(48)   Patel MK, Gacic‐Dobo M, Strebel PM, et al. Progress towards regional measles elimination – worldwide, 2000‐
2015. Weekly Epidemiological record, 45, 2016, 91, 525‐536, 11 Nov 2016. WHO: Geneva. 
(http://apps.who.int/iris/bitstream/10665/251002/1/WER9145.pdf?ua=1) 
(49)   de Martel C, Maucort‐Boulch D, Plummer M, Franceschi S. World‐wide relative contribution of hepatitis B and C 
viruses in hepatocellular carcinoma. Hepatology 2015;62:1190‐200. 
(50)   International  Agency  for  Research  on  Cancer  (2016).  HCC  cases  attributable  to  HBV  or  HCV  by  region. 
Unpublished estimates, Lyon: IARC. 
(51)   Perz JF, Armstrong GL, Farrington LA, Hutin YJ, Bell BP. The contribution of hepatitis B virus and hepatitis C virus 
infections to cirrhosis and primary liver cancer worldwide. J Hepatol. 2006; 45:529–38. 
(52)   World  Health  Organization.  The  global  burden  of  disease:  2004  update.  Geneva,  World  Health  Organization; 
2008. 
(53)    Hotez PJ, Bundy DA, Beegle K, Brooker S, Drake L, de Silva NR et al. Helminth infections: soil‐transmitted helminth 
infections and schistosomiasis. In: Jamison DT, Breman JG, Measham AR, Alleyne G, Evans D, Claeson M et al., 
eds. Disease control priorities in developing countries, 2nd edit. New York, Oxford University Press, 2006:467‐482. 
(54)   Van der Werf MJ, de Vlas SJ. Morbidity and infection with schistosomes or soil‐transmitted helminths. Rotterdam, 
Erasmus University, 2001. 

World Health Organization Page 60


 
(55)  World Health Organization. WHO Estimates of the Global Burden of Foodborne Diseases. Geneva, World Health 
Organization; 2015.  
(56)   Torgerson PR, Devleesschauwer B, Praet N, Speybroeck N, Willingham AL, et al. (2015) World Health Organization 
Estimates of the Global and Regional Disease Burden of 11 Foodborne Parasitic Diseases, 2010: A Data Synthesis. 
PLoS Med 12(12): e1001920. doi: 10.1371/journal.pmed.1001920 
(57)  Chinese Center for Disease Control and Prevention.  CDC. Reported human rabies deaths 1950‐2010. Chinese CDC: 
Beijing. 
(58)  Xx Zhou H, Vong S, Liu K, et al. Human Rabies in China, 1960‐2014: A Descriptive Epidemiological Study. 
Rupprecht CE, ed. PLoS Neglected Tropical Diseases. 2016;10(8):e0004874. doi:10.1371/journal.pntd.0004874. 
(59)  Helleringer S, Noymer A. Assessing the Direct Effects of the Ebola Outbreak on Life Expectancy in Liberia, Sierra 
Leone and Guinea. PLOS Currents Outbreaks. 2015 Feb 19 . Edition 1.  
(60)  Parpia AS, Ndeffo‐Mbah ML, Wenzel NS, et al. Effects of Response to 2014–2015 Ebola Outbreak on Deaths from 
Malaria,  HIV/AIDS,  and  Tuberculosis,  West  Africa.  Emerging  Infectious  Diseases.  2016;22(3):433‐441. 
doi:10.3201/eid2203.150977 
(61)  Takahashi S, Metcalf CJ, Ferrari MJ, Moss WJ, Truelove SA, Tatem AJ, Grenfell BT, Lessler J. Reduced vaccination 
and the risk of measles and other childhood infections post‐Ebola. Science. 2015 Mar 13;347(6227):1240‐2. doi: 
10.1126/science.aaa3438. 
(62)   Ferlay  J,  Soerjomataram  I,  Ervik  M,  Dikshit  R,  Eser  S,  Mathers  C,  Rebelo  M,  Parkin  DM,  Forman  D,  Bray,  F. 
GLOBOCAN  2012  v1.0,  Cancer  Incidence  and  Mortality  Worldwide:  IARC  CancerBase  No.  11  [Internet].  Lyon, 
International  Agency  for  Research  on  Cancer,  2013.  Available  at:  http://globocan.iarc.fr  (accessed  13  January 
2014). 
(63)  Barker  B,  Degenhardt  L.  Accidental  drug‐induced  deaths  in  Australia  1997‐2001.  NDARC  Technical  Report  No. 
163. National Drug and Alcohol Research Centre, University of New South Wales, Sydney, 2003 
(64)  Centers  for  Disease  Control  and  Prevention.  Increases  in  Drug  and  Opioid  Overdose  Deaths  —  United  States, 
2000–2014. Morbidity and Mortality Weekly Report (MMWR) 2015; 64;1‐5. 
(65)  Centers  for  Disease  Control  and  Prevention.  Prescription  Opioid  Overdose  Data.  2016. 
(http://www.cdc.gov/drugoverdose/data/overdose.html , accessed 29 August 2016). 
(66)   World Drug Report 2015. Vienna: United Nations Office on Drugs and Crime; 2015. 
(67)   Global status report on road safety 2015. Geneva, World Health Organization; 2015. 
(68)  Global status report on violence prevention 2014. Geneva, World Health Organization; 2014. 
(69)   CRED. EM‐DAT: The CRED International Disaster Database.  Louvain : Université Catholique de Louvain ; 2015.  
(70)  Uppsala Conflict Data Program. UCDP Battle‐Related Deaths Dataset v.5‐2015, 1989‐2014. Oslo, Uppsala 
University; 2015 (http://www.pcr.uu.se/research/ucdp/datasets/ucdp_battle‐related_deaths_dataset/  , 
accessed 8 July 2015). 
(71)  Uppsala Conflict Data Program. UCDP Non‐State Conflict Dataset v.5‐2015, 1989‐2014. Oslo, Uppsala University; 
2015 (http://www.pcr.uu.se/research/ucdp/datasets/ucdp_non‐state_conflict_dataset_/  , accessed 8 July 2015). 
(72)  Uppsala Conflict Data Program. UCDP One‐Sided Violence Dataset v.5‐2015, 1989‐2014. Oslo, Uppsala University; 
2015 (http://www.pcr.uu.se/research/ucdp/datasets/ucdp_one‐sided_violence_dataset/   , accessed 8 July 2015). 
(73)   Iraq Body Count. Iraqi deaths from violence 2003–2015 (http://www.iraqbodycount.org/ ) 
(74)  Iraq Family Health Survey Study Group. Violence‐Related Mortality in Iraq from 2002 to 2006. New England 
Journal of Medicine 2008, 358: 484‐93. 
(75)  Gakidou E, King G. Death by survey: estimating adult mortality without selection bias from sibling survival data. 
Demography, 2006;43:569‐585. 

World Health Organization Page 61


 
(76)  Price M, Klingner and Ball. Preliminary Statistical Analysis of Documentation of Killings in Syria. UN OHCHR 
commissioned report, January 2013 
( http://www.ohchr.org/Documents/Countries/SY/PreliminaryStatAnalysisKillingsInSyria.pdf (accessed 17 January 
2014). 
(77)  UN Secretary General Ban Ki‐moon. Statement by the Secretary‐General on Fulfilling our Collective Responsibility 
on Syria. 12 Mar 2015. http://www.un.org/sg/statements/index.asp?nid=8457 
(78)  Wikipedia. Casualties of the Syrian Civil War. (https://en.wikipedia.org/wiki/Casualties_of_the_Syrian_Civil_War , 
accessed 30 November 2015). 
(79)  Syrian Observatory for Human Rights. (http://www.syriahr.com/en/2015/02/about‐2‐millions‐killed‐and‐
wounded‐in‐47‐months‐and‐it‐is‐still‐not‐enough/ , accessed 7 January 2015). 
(80)  Office for the Coordination of Humanitarian Affairs (OCHA) ‐ Occupied Palestinian Territory (OPT). 
(http://www.ochaopt.org) 
(81)  B’Tselem – The Israeli Information Center for Human Rights in the Occupied Territories. Statistics on injuries and 
deaths suffered by both sides in the conflict. (http://www.btselem.org/statistics) 
(82)  Global terrorism database (https://www.start.umd.edu/gtd/). 
(83)  Wang H, Dwyer‐Lindgren L, Lofgren KT, et al. Age‐specific and sex‐specific mortality in 187 countries, 1970–2010: 
a systematic analysis for the Global Burden of Disease Study 2010. Lancet, 2012;380:2071–2094.  
(84)  WHO‐MCEE estimates for child causes of death 2000‐2015. Geneva: World Health Organization on behalf of 
WHO and MCEE group (http://www.who.int/healthinfo/global_burden_disease/estimates_child_cod_2015/en/ , 
accessed 15 December 2016). 

(85)  WHO. Uncertainty estimates for estimated tuberculosis deaths. 
https://extranet.who.int/tme/generateCSV.asp?ds=estimates 

(86)  WHO. Uncertainty estimates for estimated malaria deaths. 
http://www.who.int/entity/malaria/publications/world‐malaria‐report‐2016/WMR‐2016‐annexes.pdf?ua=1 

(87)  Globocan 2012: data sources and methods. Lyon: International Agency for Research on Cancer 
(http://globocan.iarc.fr/Pages/DataSource_and_methods.aspx , accessed 15 December 2016) 

(88)  World Health Organization. Health in 2015: from MDGs to SDGs. Geneva: WHO; 2015 
(http://www.who.int/gho/publications/mdgs‐sdgs/en/, accessed 9 April 2016) 

(89)  World Health Organization. World Health Statistics 2016: Monitoring health for the SDGs. Geneva: WHO; 2016 
(http://www.who.int/gho/publications/world_health_statistics/en/, accessed 20 September 2016) 

(90)  Report of the Inter‐Agency and Expert Group on Sustainable Development Goal Indicators. E/CN.3/2016/2/Rev.1. 
19 February 2016. Presented to the Forty‐seventh session of the United Nations Statistical Commission, 8–11 
March 2016 (http://unstats.un.org/unsd/statcom/47th‐session/documents/2016‐2‐IAEG‐SDGs‐Rev1‐E.pdf, 
accessed 18 April 2016). 
(91)  Report of the Commission on its 47th session. Draft decisions. 11 March 2016. New York (NY): United Nations 
Statistical Commission; 2016 (http://sd.iisd.org/events/47th‐session‐of‐un‐statistical‐commission/, accessed 9 
April 2016). 

World Health Organization Page 62


 
Annex Table A GHE cause categories and ICD‐10 codes

GHE GHE cause name ICD-10 codes
code
10 I. Communicable, maternal, perinatal and nutritional A00-B99, D50-D53, D64.9, E00-E02, E40-E46, E50-E64,
conditionsa G00-G04, G14, H65-H66, J00-J22, N70-N73, O00-O99, P00-
P96, U04
20 A. Infectious and parasitic diseases A00-B99, G00-G04, G14, N70-N73, P37.3, P37.4
30 1. Tuberculosis A15-A19, B90
40 2. STDs excluding HIV A50-A64, N70-N73
50 a. Syphilis A50-A53
60 b. Chlamydia A55-A56
70 c. Gonorrhoea A54
80 d. Trichomoniasis A59
85 e. Genital herpes A60
90 f. Other STDs A57-A58, A61-A64, N70-N73
100 3. HIV/AIDS B20-B24
101 a. HIV resulting in TB B20.0
102 b. HIV resulting in other diseases B20-B24 (minus B20.0)
110 4. Diarrhoeal diseasesb A00, A01, A03, A04, A06-A09
120 5. Childhood-cluster diseases A33-A37, B05
130 a. Whooping cough A37
140 b. Diphtheria A36
150 c. Measles B05
160 d. Tetanus A33-A35
170 6. Meningitisb A39, G00, G03
180 7. Encephalitisb A83-A86, B94.1, G04
185 8. Hepatitis B15-B19 (minus B17.8)
186 a. Acute hepatitis A B15
190 b. Acute hepatitis B B16-B19 (minus B17.1, B17.2, B18.2, B18.8)
200 c. Acute hepatitis C B17.1, B18.2
205 d. Acute hepatitis E B17.2, B18.8
210 9. Parasitic and vector diseases A71, A82, A90-A91, A95, B50-B57, B65, B67, B69, B73,
B74.0-B74.2, P37.3-P37.4

220 a. Malaria B50-B54, P37.3, P37.4


230 b. Trypanosomiasis B56
240 c. Chagas disease B57
250 d. Schistosomiasis B65
260 e. Leishmaniasis B55
270 f. Lymphatic filariasis B74.0-B74.2
280 g. Onchocerciasis B73
285 h. Cysticercosis B69
295 i. Echinococcosis B67
300 j. Dengue A90-A91
310 k. Trachoma A71
315 l. Yellow fever A95
320 m. Rabies A82
330 10. Intestinal nematode infections B76-B81
340 a. Ascariasis B77
350 b. Trichuriasis B79

World Health Organization Page 63


 
GHE GHE cause name ICD-10 codes
code

360 c. Hookworm disease B76


362 d. Food-bourne trematodes B78, B80, B81
365 11. Leprosy A30
370 12. Other infectious diseases A02, A05, A20-A28, A31, A32, A38, A40-A49, A65-A70, A74-
A79, A80-A81, A87-A89, A92-A99, B00-B04, B06-B09,
B17.8, B25-B49, B58-B60, B64, B66, B68, B70-B72, B74.3-
B74.9, B75, B82-B89, B91-B99 (minus B94.1), G14

380 B. Respiratory infectiousb H65-H66, J00-J22, P23, U04


390 1. Lower respiratory infections J09-J22, P23, U04
400 2. Upper respiratory infections J00-J06
410 3. Otitis media H65-H66
420 C. Maternal conditions O00-O99
490 D. Neonatal conditions P00-P96 (minus P23, P37.3, P37.4)
500 1. Preterm birth complicationsb P05, P07, P22, P27-P28
510 2. Birth asphyxia and birth traumab P03, P10-P15, P20-P21, P24-P26, P29
520 3. Neonatal sepsis and infections P35-P39 (minus P37.3, P37.4)
530 4. Other neonatal conditions P00-P02, P04, P08, P50-P96
540 E. Nutritional deficiencies D50-D53, D64.9, E00-E02, E40-E46, E50-E64
550 1. Protein-energy malnutrition E40-E46
560 2. Iodine deficiency E00-E02
570 3. Vitamin A deficiency E50
580 4. Iron-deficiency anaemia D50, D64.9
590 5. Other nutritional deficiencies D51-D53, E51-E64
600 II. Noncommunicable diseasesa C00-C97, D00-D48, D55-D64 (minus D 64.9), D65-D89, E03-
E07, E10-E34, E65-E88, F01-F99, G06-G98 (minus G14),
H00-H61, H68-H93, I00-I99, J30-J98, K00-K92, L00-L98,
M00-M99, N00-N64, N75-N98, Q00-Q99, X41-X42, X44,
X45, R95
610 A. Malignant neoplasms C00-C97
620 1. Mouth and oropharynx cancers C00-C14
621 a. Lip and oral cavity C00-C08
622 b. Nasopharynx C11
623 c. Other pharynx C09-C10, C12-C14
630 2. Oesophagus cancer C15
640 3. Stomach cancer C16
650 4. Colon and rectum cancers C18-C21
660 5. Liver cancerc C22
670 6. Pancreas cancer C25
680 7. Trachea, bronchus, lung cancers C33-C34
690 8. Melanoma and other skin cancers C43-C44
691 a. Malignant skin melanoma C43
692 b. Non-melanoma skin cancer C44
700 9. Breast cancer C50
710 10. Cervix uteri cancer C53
720 11. Corpus uteri cancer C54-C55
730 12. Ovary cancer C56
740 13. Prostate cancer C61
742 14. Testicular cancer C62
745 15. Kidney, renal pelvis and ureter cancer C64-C66

World Health Organization Page 64


 
GHE GHE cause name ICD-10 codes
code

750 16. Bladder cancer C67


751 17. Brain and nervous system cancers C70-C72
752 18. Gallbladder and biliary tract cancer C23-C24
753 19. Larynx cancer C32
754 20. Thyroid cancer C73
755 21. Mesothelioma C45
760 22. Lymphomas, multiple myeloma C81-C90, C96
761 a. Hodgkin lymphoma C81
762 b. Non-Hodgkin lymphoma C82-C86, C96
763 c. Multiple myeloma C88, C90
770 23. Leukaemia C91-C95
780 24. Other malignant neoplasmsd C17, C26-C31, C37-C41, C46-C49, C51, C52, C57-C60,
C63, C68, C69, C74-C80, C97

790 B. Other neoplasms D00-D48


800 C. Diabetes mellitus E10-E14 (minus E10.2-E10.29, E11.2-E11.29, E12.2, E13.2-
E13.29, E14.2)

810 D. Endocrine, blood, immune disorders D55-D64 (minus D64.9), D65-D89, E03-E07, E15-E34, E65-
E88
811 1. Thalassaemias D56
812 2. Sickle cell disorders and trait D57
813 3. Other haemoglobinopathies and haemolytic D55, D58-D59
anaemias
814 4. Other endocrine, blood and immune disorders D60-D64 (minus D64.9), D65-D89, E03-E07, E15-E34, E65-
E88
820 E. Mental and substance use disorders F04-F99, G72.1, Q86.0, X41-X42, X44, X45
830 1. Depressive disorders F32-F33, F34.1
831 a. Major depressive disorder F32-F33
832 b. Dysthymia F34.1
840 2. Bipolar disorder F30-F31
850 3. Schizophrenia F20-F29
860 4. Alcohol use disorders F10, G72.1, Q86.0, X45
e
870 5. Drug use disorders F11-F16, F18-F19e, X41-X42, X44e
871 a. Opioid use disorders F11, X42, X44e
872 b. Cocaine use disorders F14
873 c. Amphetamine use disorders F15
874 d. Cannabis use disorders F12
875 e. Other drug use disorders F13, F16, F18, F19e, X41

880 6. Anxiety disorders F40-F44


890 7. Eating disorders F50
900 8. Autism and Asperger syndrome F84
910 9. Childhood behavioural disorders F90-F92
911 a. Attention deficit/hyperactivity syndrome F90

912 b. Conduct disorder F91-F92


920 10. Idiopathic intellectual disability F70-F79
930 11. Other mental and behavioural disorders F04-F09, F17, F34-F39 (minus F34.1), F45-F48, F51-F69,
F80-F83, F88-F89, F93-F99

World Health Organization Page 65


 
GHE GHE cause name ICD-10 codes
code

940 F. Neurological conditions F01-F03, G06-G98 (minus G14, G72.1)


950 1. Alzheimer disease and other dementias F01-F03, G30-G31
960 2. Parkinson disease G20-G21
970 3. Epilepsy G40-G41
980 4. Multiple sclerosis G35
990 5. Migraine G43
1000 6. Non-migraine headache G44
1010 7. Other neurological conditions G06-G12, G23-G25, G36-G37, G45-G98 (minus G72.1)
1020 G. Sense organ diseases H00-H61, H68-H93
1030 1. Glaucoma H40
1040 2. Cataracts H25-H26
1050 3. Uncorrected refractive errors H49-H52
1060 4. Macular degeneration H35.3
1070 5. Other vision loss H30-H35 (minus H35.3), H53-H54
1080 6. Other hearing loss H90-H91
1090 7. Other sense organ disorders H00-H21, H27, H43-H47, H55-H61, H68-H83, H92-H93
1100 H. Cardiovascular diseases I00-I99
1110 1. Rheumatic heart disease I01-I09
1120 2. Hypertensive heart disease I10-I15
1130 3. Ischaemic heart diseasef I20-I25
1140 4. Strokeg I60-I69
1150 5. Cardiomyopathy, myocarditis, endocarditis I30-I33, I38, I40, I42
1160 6. Other circulatory diseases I00, I26-I28, I34-I37, I44-I51, I70-I99
1170 I. Respiratory diseases J30-J98
1180 1. Chronic obstructive pulmonary disease J40-J44
1190 2. Asthma J45-J46
1200 3. Other respiratory diseases J30-J39, J47-J98
1210 J. Digestive diseases K20-K92
1220 1. Peptic ulcer disease K25-K27
1230 2. Cirrhosis of the liverh K70, K74
1240 3. Appendicitis K35-K37
1241 4. Gastritis and duodenitis K29
1242 5. Paralytic ileus and intestinal obstruction K56
1244 6. Inflammatory bowel disease K50-K52, K58.0
1246 7. Gallbladder and biliary diseases K80-K83
1248 8. Pancreatitis K85-K86
1250 9. Other digestive diseases K20-K22, K28, K30-K31, K38, K40-K46, K55, K57, K58.9,
K59-K66, K71-K73, K75-K76, K90-K92

1260 K. Genitourinary diseases E10.2-E10.29,E11.2-E11.29,E12.2,E13.2-E13.29,E14.2,


N00-N64, N75-N76, N80-N98
1270 1. Kidney diseases N00-N19, E10.2-E10.29,E11.2-E11.29,E12.2,E13.2-
E13.29,E14.2

1271 a. Acute glomerulonephritis N00-N01


1272 b. Chronic kidney disease due to diabetes E10.2-E10.29, E11.2-E11.29, E12.2, E13.2-E13.29, E14.2

1273 c. Other chronic kidney disease N02-N19

World Health Organization Page 66


 
GHE GHE cause name ICD-10 codes
code

1280 2. Benign prostatic hyperplasia N40


1290 3. Urolithiasis N20-N23
1300 4. Other urinary diseases N25-N39, N41-N45, N47-N51
1310 5. Infertility N46, N97
1320 6. Gynecological diseases N60-N64, N75-N76, N80-N96, N98
1330 L. Skin diseases L00-L98
1340 M. Musculoskeletal diseases M00-M99
1350 1. Rheumatoid arthritis M05-M06
1360 2. Osteoarthritis M15-M19
1370 3. Gout M10
1380 4. Back and neck pain M45-M48, M50-M54
1390 5. Other musculoskeletal disorders M00, M02, M08, M11-M13, M20-M43, M60-M99
1400 N. Congenital anomalies Q00-Q99 (minus Q86.0)
1410 1. Neural tube defects Q00, Q05
1420 2. Cleft lip and cleft palate Q35-Q37
1430 3. Down syndrome Q90
1440 4. Congenital heart anomalies Q20-Q28
1450 5. Other chromosomal anomalies Q91-Q99
1460 6. Other congenital anomalies Q01-Q04, Q06-Q18, Q30-Q34, Q38-Q89 (excluding Q86.0)
1470 O. Oral conditions K00-K14
1480 1. Dental caries K02
1490 2. Periodontal disease K05
1500 3. Edentulism -
1502 4. Other oral disorders K00, K01, K03, K04, K06-K14
1505 P. Sudden infant death syndrome R95
1510 III. Injuriesi V01-Y89 (minus X41-X42, X44, X45)
1520 A. Unintentional injuries V01-X40, X43, X46-59, Y40-Y86, Y88, Y89
1530 1. Road injuryj V01-V04, V06, V09-V80, V87, V89, V99
e
1540 2. Poisonings X40, X43, X46-X48, X49e
1550 3. Falls W00-W19
1560 4. Fire, heat and hot substances X00-X19
1570 5. Drowning W65-W74
1575 6. Exposure to mechanical forces W20-W38, W40-W43, W45, W46, W49-W52, W75, W76
1580 7. Natural disasters X30-X39
1590 8. Other unintentional injuries Rest of V, W39, W44, W53-W64, W77-W99, X20-X29, X50-
X59, Y40-Y86, Y88, Y89

1600 B. Intentional injuries X60-Y09, Y35-Y36, Y870, Y871


1610 1. Self-harm X60-X84, Y870
1620 2. Interpersonal violence X85-Y09, Y871
1630 3. Collective violence and legal intervention Y35-Y36

—, not available

a
Deaths coded to “Symptoms, signs and ill-defined conditions” (R00-R94. R96-R99) are distributed proportionately to all causes within
Group I and Group II.

b
For deaths under age 5, refer to classification in Annex Table E.

World Health Organization Page 67


 
c
For liver cancer secondary to hepatitis B, hepatitis C, and alcohol use, proportions derived from GBD2013 analyses.

d
Cancer deaths coded to ICD categories for malignant neoplasms of other and unspecified sites including those whose point of origin
cannot be determined, and secondary and unspecified neoplasms (C76, C80, C97) were redistributed pro-rata across malignant neoplasm
categories within each age–sex group, so that the category “Other malignant neoplasms” includes only malignant neoplasms of other
specified sites.

e
Deaths coded to F19 (Multiple and other drug use) and X44 (Accidental poisoning by other and unspecified drugs and medicines) have
been redistributed to the GHE drug categories as described in Section 8.14. Deaths coded to X49 (Accidental poisoning by other and
unspecified chemicals) have been redistributed to GHE accidental poisoning and GHE opioid use disorders categories as described in
Section 8.14.

f
Ischaemic heart disease deaths may be miscoded to a number of so-called cardiovascular “garbage” codes. These include heart failure,
ventricular dysrhythmias, generalized atherosclerosis and ill-defined descriptions and complications of heart disease. Proportions of
deaths coded to these causes were redistributed to ischaemic heart disease as described in Mathers CD, Lopez AD, Murray CJL, Ezzati
M, Jamison DT. The burden of disease and mortality by condition: data, methods and results for 2001. Global burden of disease and risk
factors. New York, Oxford University Press, 2006. p. 45–240. Relevant ICD-10 codes are I46, I47.2, I49.0, I50, I51.4, I51.5, I51.6, I51.9
and I70.9.

g
For ischaemic stroke and haemorrhagic stroke, proportions derived from GBD2013 analyses.

h
For cirrhosis due to hepatitis B, hepatitis C, and alcohol use, proportions derived from GBD2013 analyses.

i
Injury deaths where the intent is not determined (Y10-Y34, Y872) are distributed proportionately to all causes below the group level for
injuries.

j
For countries with 3-digit ICD10 data, for “Road injury” use: V01-V04, V06, V09-V80, V87, V89 and V99. For countries with 4-digit ICD10
data, for “Road injury” use:

V01.1-V01.9, V02.1-V02.9, V03.1-V03.9, V04.1-V04.9, V06.1-V06.9, V09.2, V09.3, V10.3-V10.9, V11.3-V11.9, V12.3-V12.9, V13.3-V13.9,
V14.3-V14.9, V15.4-V15.9, V16.4-V16.9, V17.4-V17.9, V18.4-V18.9, V19.4-V19.9, V20.3-V20.9, V21.3-V21.9, V22.3-V22.9, V23.3-V23.9,
V24.3-V24.9, V25.3-V25.9, V26.3-V26.9, V27.3-V27.9, V28.3-V28.9, V29.4-V29.9, V30.4-V30.9, V31.4-V31.9, V32.4-V32.9, V33.4-V33.9,
V34.4-V34.9, V35.4-V35.9, V36.4-V36.9, V37.4-V37.9, V38.4-V38.9, V39.4-V39.9, V40.4-V40.9, V41.4-V41.9, V42.4-V42.9, V43.4-V43.9,
V44.4-V44.9, V45.4-V45.9, V46.4-V46.9, V47.4-V47.9, V48.4-V48.9, V49.4-V49.9, V50.4-V50.9, V51.4-V51.9, V52.4-V52.9, V53.4-V53.9,
V54.4-V54.9, V55.4-V55.9, V56.4-V56.9, V57.4-V57.9, V58.4-V58.9, V59.4-V59.9, V60.4-V60.9, V61.4-V61.9, V62.4-V62.9, V63.4-V63.9,
V64.4-V64.9, V65.4-V65.9, V66.4-V66.9, V67.4-V67.9, V68.4-V68.9, V69.4-V69.9, V70.4-V70.9, V71.4-V71.9, V72.4-V72.9, V73.4-V73.9,
V74.4-V74.9, V75.4-V75.9, V76.4-V76.9, V77.4-V77.9, V78.4-V78.9, V79.4-V79.9, V80.3-V80.5, V81.1, V82.1, V82.8-V82.9, V83.0-V83.3,
V84.0-V84.3, V85.0-V85.3, V86.0-V86.3, V87.0-V87.9, V89.2-V89.3, V89.9, V99 and Y850.

   

World Health Organization Page 68


 
Annex Table B Groupings of countries, areas and territories used for
global and regional tabulations
B.1 Global
 

Afghanistan, Albania, Algeria, Angola, Antigua and Barbuda, Argentina, Armenia, Australia, Austria, Azerbaijan, Bahamas, 
Bahrain, Bangladesh, Barbados, Belarus, Belgium, Belize, Benin, Bhutan, Bolivia (Plurinational State of), Bosnia and 
Herzegovina, Botswana, Brazil, Brunei Darussalam, Bulgaria, Burkina Faso, Burundi, Cabo Verde, Cambodia, Cameroon, 
Canada, Central African Republic, Chad, Chile, China; Taiwan, China; Colombia, Comoros, Congo, Costa Rica, Côte 
d'Ivoire, Croatia, Cuba, Cyprus, Czechia, Democratic People's Republic of Korea, Democratic Republic of the Congo, 
Denmark, Djibouti, Dominican Republic, Ecuador, Egypt, El Salvador, Equatorial Guinea, Eritrea, Estonia, Ethiopia, Fiji, 
Finland, France, Gabon, Gambia, Georgia, Germany, Ghana, Greece, Grenada, Guatemala, Guinea, Guinea‐Bissau, 
Guyana, Haiti, Honduras, Hungary, Iceland, India, Indonesia, Iran (Islamic Republic of), Iraq, Ireland, Israel, Italy, Jamaica, 
Japan, Jordan, Kazakhstan, Kenya, Kiribati, Kuwait, Kyrgyzstan, Lao People's Democratic Republic, Latvia, Lebanon, 
Lesotho, Liberia, Libya, Lithuania, Luxembourg, Madagascar, Malawi, Malaysia, Maldives, Mali, Malta, Mauritania, 
Mauritius, Mexico, Micronesia (Federated States of), Mongolia, Montenegro, Morocco, Mozambique, Myanmar, 
Namibia, Nepal, Netherlands, New Zealand, Nicaragua, Niger, Nigeria, Norway, Oman, Pakistan, Panama, Papua New 
Guinea, Paraguay, Peru, Philippines, Poland, Portugal, Puerto Rico, Qatar, Republic of Korea, Republic of Moldova, 
Romania, Russian Federation, Rwanda, Saint Lucia, Saint Vincent and the Grenadines, Samoa, Sao Tome and Principe, 
Saudi Arabia, Senegal, Serbia, Seychelles, Sierra Leone, Singapore, Slovakia, Slovenia, Solomon Islands, Somalia, South 
Africa, South Sudan, Spain, Sri Lanka, Sudan, Suriname, Swaziland, Sweden, Switzerland, Syrian Arab Republic, Tajikistan, 
Thailand, The former Yugoslav Republic of Macedonia, Timor‐Leste, Togo, Tonga, Trinidad and Tobago, Tunisia, Turkey, 
Turkmenistan, Uganda, Ukraine, United Arab Emirates, United Kingdom, United Republic of Tanzania, United States of 
America, Uruguay, Uzbekistan, Vanuatu, Venezuela (Bolivarian Republic of), Viet Nam, West Bank and Gaza Strip, Yemen, 
Zambia, Zimbabwe 

World Health Organization Page 69


 
B.2 WHO Region*
WHO African Region
Algeria,  Angola,  Benin,  Botswana,  Burkina  Faso,  Burundi,  Cabo  Verde,  Cameroon,  Central  African  Republic,  Chad, 
Comoros, Congo, Côte d'Ivoire, Democratic Republic of the Congo, Equatorial Guinea, Eritrea, Ethiopia, Gabon, Gambia, 
Ghana, Guinea, Guinea‐Bissau, Kenya, Lesotho, Liberia, Madagascar, Malawi, Mali, Mauritania, Mauritius, Mozambique, 
Namibia, Niger, Nigeria, Rwanda, Sao Tome and Principe, Senegal, Seychelles, Sierra Leone, South Africa, South Sudan, 
Swaziland, Togo, Uganda, United Republic of Tanzania, Zambia, Zimbabwe 

WHO Region of the Americas


Antigua  and  Barbuda,  Argentina,  Bahamas,  Barbados,  Belize,  Bolivia  (Plurinational  State  of),  Brazil,  Canada,  Chile, 
Colombia, Costa Rica, Cuba, Dominican Republic, Ecuador, El Salvador, Grenada, Guatemala, Guyana, Haiti, Honduras, 
Jamaica, Mexico, Nicaragua, Panama, Paraguay, Peru, Saint Lucia, Saint Vincent and the Grenadines, Suriname, Trinidad 
and Tobago, United States of America, Uruguay, Venezuela (Bolivarian Republic of)  

WHO South-East Asia Region


Bangladesh,  Bhutan,  Democratic  People's  Republic  of  Korea,  India,  Indonesia,  Maldives,  Myanmar,  Nepal,  Sri  Lanka, 
Thailand, Timor‐Leste 

WHO European Region


Albania,  Armenia,  Austria,  Azerbaijan,  Belarus,  Belgium,  Bosnia  and  Herzegovina,  Bulgaria,  Croatia,  Cyprus,    Czechia, 
Denmark,  Estonia,  Finland,  France,  Georgia,  Germany,  Greece,  Hungary,  Iceland,  Ireland,  Israel,  Italy,  Kazakhstan, 
Kyrgyzstan,  Latvia,  Lithuania,  Luxembourg,  Malta,  Montenegro,  Netherlands,  Norway,  Poland,  Portugal,  Republic  of 
Moldova, Romania, Russian Federation, Serbia, Slovakia, Slovenia, Spain, Sweden, Switzerland, Tajikistan,  The former 
Yugoslav Republic of Macedonia, Turkey, Turkmenistan, Ukraine, United Kingdom, Uzbekistan 

WHO Eastern Mediterranean Region


Afghanistan, Bahrain, Djibouti, Egypt, Iran (Islamic Republic of), Iraq, Jordan, Kuwait, Lebanon, Libya, Morocco, Oman, 
Pakistan, Qatar, Saudi Arabia, Somalia, Sudan, Syrian Arab Republic, Tunisia, United Arab Emirates, Yemen 
WHO Western Pacific Region
Australia,  Brunei  Darussalam,  Cambodia,  China,  Fiji,  Japan,  Kiribati,  Lao  People's  Democratic  Republic,  Malaysia, 
Marshall Islands, Micronesia (Federated States of), Mongolia, New Zealand, Papua New Guinea, Philippines, Republic of 
Korea, Samoa, Singapore, Solomon Islands, Tonga, Vanuatu, Viet Nam 
 
 
*WHO regional grouping as of 2015. WHO Member States with a population of less than 90 000 population in 2015 were not included 
in the analysis; these include: Andorra, Cook Islands, Dominica, Marshall Islands, Monaco, Nauru, Niue, Palau, Saint Kitts and Nevis, 
San Marino, Tuvalu.    

 

World Health Organization Page 70


 
B.3 World Bank income grouping*
Low income
Afghanistan,  Benin,  Burkina  Faso,  Burundi,  Central  African  Republic,  Chad,  Comoros,  Democratic  People's  Republic  of 
Korea, Democratic Republic of the Congo, Eritrea, Ethiopia, Gambia, Guinea, Guinea‐Bissau, Haiti, Liberia, Madagascar, 
Malawi, Mali, Mozambique, Nepal, Niger, Rwanda, Senegal, Sierra Leone, Somalia, South Sudan, Togo, Uganda, United 
Republic of Tanzania, Zimbabwe 

Lower middle income


Armenia, Bangladesh, Bhutan, Bolivia (Plurinational State of), Cabo Verde, Cambodia, Cameroon, Congo, Côte d'Ivoire, 
Djibouti,  Egypt,  El  Salvador,  Ghana,  Guatemala,  Honduras,  India,  Indonesia,  Kenya,  Kiribati,  Kyrgyzstan,  Lao  People's 
Democratic Republic, Lesotho, Mauritania, Micronesia (Federated States of), Mongolia, Morocco, Myanmar, Nicaragua, 
Nigeria,  Pakistan,  Papua  New  Guinea,  Philippines,  Republic  of  Moldova,  Samoa,  Sao  Tome  and  Principe,  Solomon 
Islands, Sri Lanka, Sudan, Swaziland, Syrian Arab Republic, Tajikistan, Timor‐Leste, Tonga, Tunisia, Ukraine, Uzbekistan, 
Vanuatu, Viet Nam, West Bank and Gaza Strip, Yemen, Zambia 

Upper middle income


Albania,  Algeria,  Angola,  Argentina,  Azerbaijan,  Belarus,  Belize,  Bosnia  and  Herzegovina,  Botswana,  Brazil,  Bulgaria,  
China,  Colombia,  Costa  Rica,  Cuba,  Dominican  Republic,  Ecuador,  Equatorial  Guinea,  Fiji,  Gabon,  Georgia,  Grenada, 
Guyana,  Iran  (Islamic  Republic  of),  Iraq,  Jamaica,  Jordan,  Kazakhstan,  Lebanon,  Libya,  Malaysia,  Maldives,  Mauritius, 
Mexico, Montenegro, Namibia, Panama, Paraguay,  Peru, Romania, Russian Federation, Saint Lucia, Saint Vincent and 
the  Grenadines,  Serbia,  South  Africa,  Suriname,  Thailand,  The  former  Yugoslav  Republic  of  Macedonia,  Turkey, 
Turkmenistan, Venezuela (Bolivarian Republic of) 

High income
Antigua  and  Barbuda,  Australia,  Austria,  Bahamas,  Bahrain,  Barbados,  Belgium,  Brunei  Darussalam,  Canada,  Chile, 
Taiwan, China; Croatia, Cyprus,  Czechia, Denmark, Estonia, Finland, France, Hungary, Germany, Greece, Iceland, Ireland, 
Israel,  Italy,  Japan,  Kuwait,  Latvia,  Lithuania,  Luxembourg,  Malta,  Netherlands,  New  Zealand,  Norway,  Oman,  Poland, 
Portugal, Puerto Rico, Qatar, Republic of Korea, Saudi Arabia, Seychelles, Singapore, Slovakia, Slovenia, Spain, Sweden, 
Switzerland, Trinidad and Tobago, United Arab Emirates, United Kingdom, United States of America, Uruguay 
 
* This regional grouping classifies countries, areas and territories according to the World Bank analytical income of economies  based 
on the 2015 Atlas gross national income per capita estimates (World Bank list of economies, July 2016). 

World Health Organization Page 71


 
B.4 World Bank regions
East Asia and Pacific
Australia, Brunei Darussalam, Cambodia, China; Taiwan, China; Democratic People's Republic of Korea, Fiji, Indonesia, 
Japan, Kiribati, Lao People's Democratic Republic, Malaysia, Micronesia (Federated States of), Mongolia, Myanmar, New 
Zealand, Papua New Guinea, Philippines, Republic of Korea, Samoa, Singapore, Solomon Islands, Thailand, Timor‐Leste, 
Tonga, Vanuatu, Viet Nam 

Europe and Central Asia


Albania,  Armenia,  Austria,  Azerbaijan,  Belarus,  Belgium,  Bosnia  and  Herzegovina,  Bulgaria,  Croatia,  Cyprus,    Czechia, 
Denmark, Estonia, Finland, France, Georgia, Germany, Greece, Hungary, Iceland, Ireland, Italy, Kazakhstan, Kyrgyzstan, 
Latvia,  Lithuania,  Luxembourg,  Montenegro,  Netherlands,  Norway,  Poland,  Portugal,  Republic  of  Moldova,  Romania, 
Russian Federation, Serbia, Slovakia, Slovenia, Spain, Sweden, Switzerland, Tajikistan, The former Yugoslav Republic of 
Macedonia, Turkey, Turkmenistan, Ukraine, United Kingdom, Uzbekistan 

Latin America and Caribbean


Antigua  and  Barbuda,  Argentina,  Bahamas,  Barbados,  Belize,  Bolivia  (Plurinational  State  of),  Brazil,  Chile,  Colombia, 
Costa  Rica,  Cuba,  Dominican  Republic,  Ecuador,  El  Salvador,  Grenada,  Guatemala,  Guyana,  Haiti,  Honduras,  Jamaica, 
Mexico,  Nicaragua,  Panama,  Paraguay,  Peru,  Puerto  Rico,  Saint  Lucia,  Saint  Vincent  and  the  Grenadines,  Suriname, 
Trinidad and Tobago, Uruguay, Venezuela (Bolivarian Republic of) 

Middle East and North Africa


Algeria, Bahrain, Djibouti, Egypt, Iran (Islamic Republic of), Iraq, Israel, Jordan, Kuwait, Lebanon, Libya, Malta, Morocco, 
Oman, Qatar, Saudi Arabia, Syrian Arab Republic, Tunisia, United Arab Emirates, West Bank and Gaza Strip, Yemen 

North America

Canada, United States of America 

South Asia
Afghanistan, Bangladesh, Bhutan, India, Maldives, Nepal, Pakistan, Sri Lanka 

Sub-Saharan Africa
Angola,  Benin,  Botswana,  Burkina  Faso,  Burundi,  Cabo  Verde,  Cameroon,  Central  African  Republic,  Chad,  Comoros, 
Congo, Côte  d'Ivoire,  Democratic  Republic  of  the  Congo,  Equatorial  Guinea,  Eritrea,  Ethiopia,  Gabon,  Gambia,  Ghana, 
Guinea,  Guinea‐Bissau,  Kenya,  Lesotho,  Liberia,  Madagascar,  Malawi,  Mali,  Mauritania,  Mauritius,  Mozambique, 
Namibia, Niger, Nigeria, Rwanda, Sao Tome and Principe, Senegal, Seychelles, Sierra Leone, Somalia, South Africa, South 
Sudan, Sudan, Swaziland, Togo, Uganda, United Republic of Tanzania, Zambia, Zimbabwe  
 
   

World Health Organization Page 72


 
Annex Table C GATHER checklist
 
Item # Checklist item Location reported
Objectives and funding
1 Define  the  indicator(s),  populations  (including  age,  sex,  and  geographic  Sections 2‐3 
entities), and time period(s) for which estimates were made. 
2 List the funding sources for the work.  Acknowledgments 
Data Inputs
For all data inputs from multiple sources that are synthesized as part of the study:
3 Describe how the data were identified and how the data were accessed.   Section 4.1 
4 Specify the inclusion and exclusion criteria. Identify all ad‐hoc exclusions.  Section 4.2 
5 Provide  information  on  all  included  data  sources  and  their  main  Table  4.1:  data  with 
characteristics. For each data source used, report reference information or  “Excluded”  in  the  notes 
contact name/institution, population represented, data collection method,  column were not used 
year(s)  of  data  collection,  sex  and  age  range,  diagnostic  criteria  or 
measurement method, and sample size, as relevant.  
6 Identify  and  describe  any  categories  of  input  data  that  have  potentially  N/A 
important biases (e.g., based on characteristics listed in item 5). 
For data inputs that contribute to the analysis but were not synthesized as part of the study:
7 Describe and give sources for any other data inputs.    
Population by age and sex  Section 2 
Total number of deaths by age and sex  Section 2 
China/India  Sections 6‐7 
Program estimates of cause of death  Section 8 
GBD2015 estimates for causes of death  Section 9 
For all data inputs:
8 Provide  all  data  inputs  in  a  file  format  from  which  data  can  be  efficiently  http://www.who.int/healt
extracted  (e.g.,  a  spreadsheet  rather  than  a  PDF),  including  all  relevant  hinfo/global_burden_dise
meta‐data  listed  in  item  5.  For  any  data  inputs  that  cannot  be  shared  ase/estimates/en/index1.
because of ethical or legal reasons, such as third‐party ownership, provide  html 
a contact name or the name of the institution that retains the right to the   
data. 
Data analysis
9 Provide a conceptual overview of the data analysis method. A diagram may  Section 1 
be helpful.  
10 Provide  a  detailed  description  of  all  steps  of  the  analysis,  including  Sections 4‐10 
mathematical  formulae.  This  description  should  cover,  as  relevant,  data 
cleaning,  data  pre‐processing,  data  adjustments  and  weighting  of  data 
sources, and mathematical or statistical model(s).  
11 Describe how candidate models were evaluated and how the final model(s)  N/A:  statistical  models 

World Health Organization Page 73


 
were selected.  were  not  used  to 
synthesize data 
12 Provide the results of an evaluation of model performance, if done, as well  N/A:  statistical  models 
as the results of any relevant sensitivity analysis.  were  not  used  to 
synthesize data 
13 Describe methods for calculating uncertainty of the estimates. State which  Section 10 
sources  of  uncertainty  were,  and  were  not,  accounted  for  in  the 
uncertainty analysis. 
14 State how analytic or statistical source code used to generate estimates can  Acknowledgments 
be accessed.  (available  upon  request 
from [email protected]
Results and Discussion
15 Provide  published  estimates  in  a  file  format  from  which  data  can  be  http://www.who.int/healt
efficiently extracted.  hinfo/global_burden_dise
ase/estimates/en/index1.
html 
 
16 Report  a  quantitative  measure  of  the  uncertainty  of  the  estimates  (e.g.  Section 10 
uncertainty intervals). 
17 Interpret results in light of existing evidence. If updating a previous set of  Section 11, Annex Table D 
estimates, describe the reasons for changes in estimates. 
18 Discuss limitations of the estimates. Include a discussion of any modelling  Section 11 
assumptions or data limitations that affect interpretation of the estimates. 

World Health Organization Page 74


 
Annex Table D Methods used for estimation of mortality levels
and causes of death, by country, 2000‐2015
 
All‐cause mortality method groups:
A:   Life tables based on death rates computed from vital registration data.   
B:   Life tables based on UNPD’s World Population Prospects – the 2015 revision, and child mortality estimates from 
the UN‐IGME.   
C:   Life tables based on UNPD’s World Population Prospects – the 2015 revision, updated with the latest HIV/AIDS 
mortality from UNAIDS and child mortality estimates from the UN‐IGME 
D:   WHO modelled HIV and non‐HIV mortality. 

Abbreviations
GBD2015   Global Burden of Disease 2015 study estimates (6) 
GBD2015  GBD2015 study estimates drawing on WHO death registration data (VR) for the country 
High HIV  WHO/UNAIDS/WPP2015 estimates for HIV deaths and all‐cause deaths, GBD2015 study estimates (6) 
VA/VR     Verbal autopsy and Verbal autopsy sample data plus sample death registration data 
VR      Vital (death) registration 
Note (a)  WHO and UN Interagency cause‐specific estimates (see Section 8 above). 
           n.a.  Useability not assessed 
 

Average 
All‐cause  Under 5 child  Latest  useability 
mortality  cause of death  Cause of death  available  2000‐
Country  method  method  methods for ages 5+  year VR  latest 
Afghanistan  B  VA MCM  GBD2015 plus (a) 
Albania  A  VR MCM  GBD2015  2009  60% 
Algeria  B  VA MCM  GBD2015 plus (a) 
Angola  C  VA MCM  High HIV 
Antigua and Barbuda  B  VR data  VR data  2014  78% 
Argentina  A  VR data  GBD2015  2014  77% 
Armenia  A  VR MCM  GBD2015  2015  95% 
Australia  A  VR data  VR data  2014  95% 
Austria  A  VR data  VR data  2014  91% 
Azerbaijan  A  VA MCM  GBD2015  2007  87% 
Bahamas  D  VR data  VR data & High HIV  2013  87% 
Bahrain  B  VR data  GBD2015  2014  61% 
Bangladesh  B  VA MCM  GBD2015 plus (a) 
   

World Health Organization Page 75


 
        Average 
  All‐cause  Under 5 child    Latest  useability 
  mortality  cause of death  Cause of death  available  2000‐
Country  method  method  methods for ages 5+  year VR  latest 
68% 
Barbados  A  VR data  VR data  2013   
 
Belarus  A  VR MCM  VR data  2014  80% 
Belgium  A  VR data  VR data  2013  87% 
Belize  D  VR data  VR data & High HIV  2014  76% 
Benin  D  VA MCM  High HIV 
Bhutan  B  VA MCM  GBD2015 plus (a) 
Bolivia (Plurinational State of)  A  VA MCM  GBD2015  2003  17% 
Bosnia and Herzegovina  A  VR MCM  GBD2015  2014  73% 
Botswana  C  VA MCM  High HIV 
Brazil  A  VR data  VR data  2014  83% 
Brunei Darussalam  A  VR data  VR data  2014  91% 
Bulgaria  A  VR data  GBD2015  2013  76% 
Burkina Faso  D  VA MCM  High HIV 
Burundi  C  VA MCM  High HIV 
Cabo Verde  B  VR MCM  GBD2015  2011  73% 
Cambodia  B  VA MCM  GBD2015 plus (a) 
Cameroon  C  VA MCM  High HIV 
VR MCM (0‐
27d), VR data 
Canada  A  (1‐59m)   VR data  2012  94% 
Central African Republic  C  VA MCM  High HIV 
Chad  D  VA MCM  High HIV 
Chile  A  VR data  VR data  2014  94% 
China  B  Sample VR  VA/VR data   2013  n.a. 
Colombia  A  VR data  VR data  2013  84% 
Comoros  B  VA MCM  GBD2015 plus (a) 
Congo  C  VA MCM  High HIV 
Costa Rica  A  VR data  VR data  2014  87% 
Côte d'Ivoire  D  VA MCM  High HIV 
Croatia  A  VR data  VR data  2015  89% 
Cuba  A  VR data  VR data  2014  92% 
Cyprus  B  VR MCM  VR data  2013  58% 
Czechia  A  VR data  VR data  2015  89% 
Democratic People's Republic of 
Korea  B  VA MCM  GBD2015 plus (a) 
Democratic Republic of the Congo  D  VA MCM  High HIV 
Denmark  A  VR data  VR data  2014  87% 
Djibouti  D  VA MCM  High HIV 
Dominican Republic  A  VA MCM  GBD2015  2012  45% 
Ecuador  A  VR MCM  VR data  2014  69% 

World Health Organization Page 76


 
Average 
All‐cause  Under 5 child  Latest  useability 
mortality  cause of death  Cause of death  available  2000‐
Country  method  method  methods for ages 5+  year VR  latest 
 
Egypt  A  VR MCM  GBD2015  2011  56% 
El Salvador  A  VR MCM  GBD2015  2013  63% 
Equatorial Guinea  C  VA MCM  High HIV 
Eritrea  D  VA MCM  High HIV 
Estonia  A  VR data  VR data  2014  94% 
Ethiopia  C  VA MCM  High HIV 
Fiji  B  VR MCM  GBD2015  2012  81% 
Finland  A  VR data  VR data  2014  97% 
France  A  VR data  VR data  2013  85% 
Gabon  C  VA MCM  High HIV 
Gambia  D  VA MCM  High HIV 
Georgia  A  VR MCM  GBD2015  2014  61% 
Germany  A  VR data  VR data  2014  88% 
Ghana  D  VA MCM  High HIV  2014  n.a. 
Greece  A  VR data  GBD2015  2013  74% 
Grenada  B  VR data  VR data  2015  88% 
Guatemala  A  VA MCM  VR data  2014  79% 
Guinea  D  VA MCM  High HIV 
Guinea‐Bissau  D  VA MCM  High HIV 
Guyana  A  VR data  VR data  2012  86% 
Haiti  D  VA MCM  High HIV  2004  6% 
Honduras  B  VR MCM  GBD2015  2013  13% 
Hungary  A  VR data  VR data  2014  95% 
Iceland  A  VR data  VR data  2015  94% 
India  B  State level  VA/VR data   2007  n.a. 
Indonesia  B  VA MCM  GBD2015 plus (a) 
Iran (Islamic Republic of)  B  VA MCM  GBD2015 plus (a)  2014  n.a. 
Iraq  B  VA MCM  GBD2015  2008  54% 
Ireland  A  VR data  VR data  2013  95% 
Israel  A  VR data  VR data  2014  91% 
Italy  A  VR data  VR data  2012  91% 
Jamaica  D  VR MCM  VR data & High HIV  2011  73% 
Japan  A  VR data  VR data  2014  89% 
Jordan  B  VR MCM  GBD2015  2011  61% 
Kazakhstan  A  VA MCM  VR data  2015  81% 
Kenya  C  VA MCM  High HIV  2015  n.a. 
Kiribati  B  VA MCM  GBD2015  2001  51% 
Kuwait  B  VR data  GBD2015  2014  65% 
Kyrgyzstan  A  VA MCM  VR data  2015  90% 

World Health Organization Page 77


 
Average 
All‐cause  Under 5 child  Latest  useability 
mortality  cause of death  Cause of death  available  2000‐
Country  method  method  methods for ages 5+  year VR  latest 
Lao People's Democratic Republic  B  VA MCM  GBD2015 plus (a) 
Latvia  A  VR data  VR data  2014  90% 
Lebanon  B  VR MCM  GBD2015 plus (a)  1999  n.a. 
Lesotho  C  VA MCM  High HIV 
Liberia  D  VA MCM  High HIV 
Libya  B  VR MCM  GBD2015 plus (a) 
Lithuania  A  VR data  VR data  2015  92% 
Luxembourg  A  VR data  VR data  2014  86% 
Madagascar  B  VA MCM  GBD2015 plus (a)  2010  n.a. 
Malawi  C  VA MCM  High HIV 
Malaysia  B  VR MCM  GBD2015  2008  40% 
Maldives  A  VR MCM  GBD2015  2011  58% 
Mali  D  VA MCM  High HIV 
Malta  A  VR data  VR data  2014  92% 
Mauritania  B  VA MCM  GBD2015 plus (a) 
Mauritius  A  VR data  VR data  2014  88% 
Mexico  A  VR data  VR data  2014  95% 
Micronesia (Federated States of)  B  VA MCM  GBD2015 plus (a) 
Mongolia  A  VA MCM  GBD2015 plus (a)  2012  n.a. 
Montenegro  A  VR data  GBD2015  2009  67% 
Morocco  B  VA MCM  GBD2015  2012  12% 
Mozambique  C  VA MCM  High HIV  2014  n.a. 
Myanmar  B  VA MCM  GBD2015 plus (a)  2014  n.a. 
Namibia  C  VA MCM  High HIV 
Nepal  B  VA MCM  GBD2015 plus (a) 
Netherlands  A  VR data  VR data  2015  86% 
New Zealand  A  VR data  VR data  2012  97% 
Nicaragua  A  VR MCM  GBD2015  2013  60% 
Niger  B  VA MCM  GBD2015 plus (a) 
Nigeria  D  VA MCM  High HIV 
Norway  A  VR data  VR data  2014  88% 
Oman  B  VR MCM  GBD2015  2010  49% 
Pakistan  B  VA MCM  GBD2015 plus (a) 
Panama  A  VR data  VR data  2014  80% 
Papua New Guinea  B  VA MCM  GBD2015 plus (a) 
Paraguay  B  VR MCM  GBD2015  2014  61% 
Peru  A  VR MCM  GBD2015  2014  57% 
Philippines  A  VA MCM  VR data  2011  78% 
Poland  A  VR data  GBD2015  2014  72% 

World Health Organization Page 78


 
Average 
All‐cause  Under 5 child  Latest  useability 
mortality  cause of death  Cause of death  available  2000‐
Country  method  method  methods for ages 5+  year VR  latest 
VR MCM (0‐
27d), VR data 
Portugal  A  (1‐59m)   VR data  2014  82% 
Qatar  B  VR MCM  GBD2015  2012  60% 
Republic of Korea  A  VR data  VR data  2013  85% 
Republic of Moldova  A  VR data  VR data  2015  83% 
Romania  A  VR data  VR data  2015  92% 
Russian Federation  A  VR MCM  VR data  2011  89% 
Rwanda  C  VA MCM  High HIV 
Saint Lucia  A  VR data  VR data  2014  75% 
Saint Vincent and the Grenadines  B  VR data  VR data  2015  93% 
Samoa  B  VR MCM  GBD2015 plus (a) 
Sao Tome and Principe  B  VA MCM  GBD2015 plus (a) 
Saudi Arabia  B  VR MCM  GBD2015  2012  21% 
Senegal  B  VA MCM  GBD2015 plus (a) 
Serbia  A  VR data  VR data  2014  80% 
Seychelles  B  VR MCM  GBD2015 plus (a)  2015  n.a. 
Sierra Leone  D  VA MCM  High HIV 
Singapore  B  VR data  VR data  2015  71% 
Slovakia  A  VR data  VR data  2014  94% 
Slovenia  A  VR data  VR data  2015  89% 
Solomon Islands  B  VA MCM  GBD2015 plus (a) 
Somalia  B  VA MCM  GBD2015 plus (a) 
VR data (0‐27d), 
VA  MCM (1‐
South Africa  C  59m)  High HIV  2014  70% 
South Sudan  D  VA MCM  High HIV 
Spain  A  VR data  VR data  2014  90% 
Sri Lanka  B  VR MCM  GBD2015  2006  72% 
Sudan  B  VA MCM  GBD2015 plus (a)  2010  n.a. 
Suriname  A  VR data  GBD2015  2014  64% 
Swaziland  C  VA MCM  High HIV 
Sweden  A  VR data  VR data  2015  89% 
VR MCM (0‐
27d), VR data 
Switzerland  A  (1‐59m)   VR data  2013  89% 
Syrian Arab Republic  B  VR MCM  GBD2015  2010  72% 
Tajikistan  A  VA MCM  GBD2015  2005  78% 
Thailand  D  VR MCM  High HIV  2014  52% 
The former Yugoslav Republic of 
Macedonia  A  VR data  VR data  2013  87% 
Timor‐Leste  B  VA MCM  GBD2015 plus (a) 

World Health Organization Page 79


 
Average 
All‐cause  Under 5 child  Latest  useability 
mortality  cause of death  Cause of death  available  2000‐
Country  method  method  methods for ages 5+  year VR  latest 
Togo  D  VA MCM  High HIV 
Tonga  B  VR MCM  GBD2015 plus (a) 
Trinidad and Tobago  A  VR data  VR data  2010  84% 
Tunisia  B  VR MCM  GBD2015  2013  22% 
Turkey  B  VR data  GBD2015  2013  58% 
Turkmenistan  A  VA MCM  GBD2015  2013  67% 
Uganda  C  VA MCM  High HIV 
Ukraine  A  VR MCM  VR data  2014  90% 
United Arab Emirates  B  VR MCM  GBD2015  2010  56% 
United Kingdom  A  VR data  VR data  2014  93% 
United Republic of Tanzania  C  VA MCM  High HIV  2014  n.a. 
United States of America  A  VR data  VR data  2014  93% 
Uruguay  A  VR data  VR data  2014  83% 
Uzbekistan  A  VA MCM  VR data  2014  85% 
Vanuatu  B  VR MCM  GBD2015 plus (a) 
Venezuela (Bolivarian Republic of)  A  VR data  VR data  2013  82% 
Viet Nam  B  VR MCM  GBD2015 plus (a) 
Yemen  B  VA MCM  GBD2015 plus (a) 
Zambia  C  VA MCM  High HIV  2014  n.a. 
Zimbabwe  C  VA MCM  High HIV  2002  n.a. 

 
 

World Health Organization Page 80


 
Annex Table E First‐level categories for analysis of child causes of death

GBD cause name ICD-10 code ICD-9 code

All causes A00-Y89 001-999

I. Communicable, maternal, A00-B99, D50-D53, D64.9, E00-E02, 001-139, 243, 260-269, 279.5-279.6,
perinatal and nutritional E40-E64, G00-G09, H65-H66, J00- 280, 281, 285.9, 320-326, 381-382,
conditions
a J22, J85, N30, N34, N390, N70-N73, 460-466, 480-487, 513, 614-616, 630-
O00-P96, U04 676, 760-779

HIV/AIDS B20-B24 279.5-279.6, 042

Diarrhoeal diseases A00-A09 001-009

Pertussis A37 033

Tetanus A33-A35 037, 771.3

Measles B05 055

Meningitis/encephalitis A20.3, A32.1, A39.1, G00–G09 036, 320, 322-326

Malaria B50-B54, P37.3, P37.4 084

Acute respiratory infections H65-H66, J00-J22, J85, P23, U04 460-466, 480-487, 381-382, 513,
770.0

Prematurity P01.0, P01.1, P07, P22, P25-P28, 761.0-761.1, 765, 769, 770.2-770.9,
P52, P61.2, P77 772.1, 774.2, 776.6, 777.5-777.6,

b
Birth asphyxia & birth trauma P01.7-P02.1, P02.4-P02.6, P03, P10- 761.7-762.1, 762.4-762.6, 763, 767-
P15, P20-P21, P24, P50, P90-P91 768, 770.1, 772.2, 779.0-779.2

Sepsis and other infectious P35-P39 (exclude P37.3, P37.4)


conditions of the newborn 771.0-771.2, 771.4-771.8

Other Group I Remainder Remainder

II. Noncommunicable C00-C97, D00-D48, D55-D64 (exclude 140- 242, 244-259, 270-279, 282-285,
a
diseases D64.9), D65-D89, E03-E34, E65-E88, 286-319, 330-380, 383-459, 470-478,
F01-F99, G10-G98, H00-H61, H68- 490- 512, 514-611, 617- 629, 680- 759
H93, I00-I99, J30-J84, J86-J98, K00- (exclude 279.5-279.6, 285.9)
K92, L00-L98, M00-M99, N00-N28,
N31-N32, N35-N64 (exclude N39.0),
N75-N98, Q00-Q99

Congenital anomalies Q00-Q99 740-759

Other Group II Remainder Remainder

III. Injuries V01-Y89 E800-E999


a
Deaths coded to “Symptoms, signs and ill-defined conditions” (780-799 in ICD-9 and R00-R99 in ICD-10) are distributed proportionately to all for neonatal
deaths, but exclusively to Group I and Group II for the postneonatal deaths.
b
Also referred to as “intrapartum-related complications”

World Health Organization Page 81


 

You might also like