GHME 2013 Conference Session: Global Burden of Diseases, Injuries, and Risk Factors Study 2010: workshop on methods and key findings Date: June 18 2013 Presenter: Rafael Lozano Institute: Institute for Health Metrics and Evaluation (IHME), University of Washington

1. Causes of death
Rafael Lozano, MD.
Director of the Center for Health System Research, NIPH, Mexico
Director of the Latin America and the Caribbean initiatives, IHME.
Global Burden of Diseases, Injuries, and Risk Factors Study
2010: workshop on methods and key findings
June 18, 2013

2. Outline
• Background
• Methods: Main Issues for calculating causes of death
• Key findings
• Conclusions
2

3. Background
• Causes of death (CoD) is one of the most fundamental metrics
for population health.
• Trends in CoD provide an important summary of whether
society is or is not making progress in reducing burden of
premature mortality and especially avoidable mortality.
• Usually CoD assessments show success and failures of Health
Information Systems and provide directions of how to improve
them.
• GBD 1990 was the first comprehensive study to present the
global leading causes of death.
3

4. 4

5. Global cause of death assessment: main
issues
• The universe of data
• Efforts to assess and enhance quality and comparability
of data
• The statistical modeling strategy
• Causes of death constrained to sum to all cause mortality
5

6. The universe of CoD data
We have to identify all available
data on causes of death for all
countries for a period of time
We have to identify all different
sources of CoD data
• We collected data on around 600
million deaths in the last 30 years
• Data available varies by disease:
o More on maternal, cancer, injuries
o Less on NTD, diarrhea and LRI
pathogens
6
Only VR
All Sources

7. Data Sources, Countries, and Site years, 1900-2010
7

8. Assessment and enhancement of data
quality and comparability
1. Assessment of
completeness
2. Causes of death
mapping
3. Redistribution of
misclassified causes of
death
4. Age and age-sex
splitting
5. Smoothing for
stochastic variation due
to small numbers
6. Outlier detection
8

9. Mapping one cause in different ICD
9

10. Mapping the cause of death lists
10
BTL1 2 3 4 5 Tab B 6,7 Tab A 8
9 tab
9 VA 10
Tab
10
Different type of ICD
1900 2010
SPC
countries
map
Different type of ICD
GBD 1990 Cause List (100)
GBD 2010 Cause List (235)
24 causes
39 causes

11. 11
Group A: Communicable, maternal, perinatal and nutritional conditions
D. Intestinal infectious diseases
1. Diarrheal disease
Group B: Non communicable diseases
H. Cardiovascular and circulatory diseases
Group C: Injuries
A. Unintentional injuries
a .Cholera
c. Shigellosis
i. Rotaviral enteritis
4. Cerebrovascular disease
b. Hemorrhagic stroke
1. Transport Injuries
a. Road injury
a3. injury - motorized two-wheeler rider
GBD2010 has a list of 235 CoD
Using ICD9th and ICD10th revisions
Top down hierarchical map

12. 12
Percent of Garbage Codes by
type, and sex in all ICD10 ,IHME
data set (26.6%)

13. 13

14. Garbage Codes from ICD10
(Percent of deaths
14

15. Redistribution of Garbage Codes
15
Garbage Codes
C26 ill-defined Cancers for digestive organs
Target Codes
C16 Stomach cancer
C17 Cancer of the small intestine
C18 Colon cancer
C19 Malignant neoplasm of rectosigmoid junction
C23 Gallbladder cancer
C24 Malignant neoplasm of other parts of biliary
tract
C25 Pancreatic cancer

16. Modeling causes of death
1. Causes of death ensemble modeling, CODEm (133 causes),
including all major causes except HIV. CODEm selects
models and ensembles of models based on out-of-sample
performance.
2. Negative binomial (12 causes).
3. Fixed proportion models (27 causes).
4. Disaggregation by pathogens or sub-causes (36 causes).
5. Natural history models (8 causes).
6. Mortality shock regressions (2 causes).
16

17. Combining results: CoDCorrect algorithm
• Because we developed single-cause models, it was imperative
as a final step to ensure that individual cause estimates
summed to the all-cause mortality estimate for every age-sex-
country-year group.
• This is one of the innovations of this study:
o Implemented taking into account uncertainty in every cause of
death model outcome
o We proportionately rescaled every cause such that the sum of the
cause-specific estimates equaled the number of deaths from all
causes generated from the demographic analysis (by country,
year, age, and sex).
o We applied CoDCorrect in a hierarchical way.
17

18. Shift of causes of death in the last 20 years
18
Comm/Mater/
Neonat/Nutr
34%
Injuries
9%
Comm/Mater
/Neonat/Nutr
25%
Injuries
10%
1990 2010
Non Communicable
Diseases
57%
Non Communicable
Diseases
65%
46.5 million 52.7 million

19. Death decomposition analysis by changes in
population
19
% change 1990-2000 % change due to change in rates
% change due to pop ageing % change due to pop growth
-75%
-50%
-25%
0%
25%
50%
75%
All Causes Com/Mat
Neo/Nut
NCD Injuries

20. Percentage of global deaths for female and
male individuals in 2010 by cause and age
Males Females

21. 21
Rapid
shifts in
leading
causes of
global
death

22. Percentage of YLLs for all ages and both sexes
combined by cause and region in 1990 and 2010
1990 2010

23. 23

24. Key findings
• The shifting pattern of the number of deaths by cause across
time, countries, and age groups is consistent with the three key
drivers of change
• Despite the important epidemiological shift in the world, the
MDGs related deaths in Sub Sahara Africa represent 60% of
all deaths in that region during 2010
• New set of analytical approaches and methods:
o Improved diagnostic redistribution
o The modeling strategy depends of the strength of available data:
CODEM and CoD Correct are both innovations in the field
• Adding time trends and quantifying the uncertainty differentiate
GBD 2010 from similar studies in the past, however without
correction of known bias, comparability is impossible.
24