Global Burden of Disease Arindam Basu, 4 th March, 2013
Global Burden of Disease
Arindam Basu, 4th March, 2013
Concepts and models
Who Enjoys Most Health?
Ann, 34, F
Bob, 60, M
Tom, 40, M
Jan, 80, F
20072012
Full Health Disease/Injury Disability
(Year of their death)
How Can We Bring Diseases and Deaths into One Common Rubric?
Why Should We Need to do that?
Problems of Reliance on Standard Epidemiological Data
• Aggregate Measures can become unwieldy• Many Problems of validity• Not good for comparisons• Partial and fragmented health statistics• Risk of Overestimation• Separate measures for death and
incidence/prevalence
What is GBD
• GBD is a comprehensive effort to estimate these summary measures
• GBD uses Disability Adjusted Life Years• DALY = YLL + YLD• Disability == Health Loss, other than death• Seven institutions - IIHME, UQueensland,
WHO, Hopkins, Harvard, Imperial College, Tokyo Uni
Why Do We Need GBD?
• Need to Understand what is the “problem”, or “big picture”?
• What are the problems? Are they getting better or worse?• Combine this with effectiveness, equity, for policy• Compare overall population health across communities• Compare over time• Get a coherent overall picture of diseases that contribute
most to loss of health• What are our best data sources?• What data sources are high quality sources?
Life Expectancy, YLL, and YLD
Birth Death
(Life Expectancy At Birth)
Life Expectancy At Birth
(Years of Life Lost)
Life Expectancy At Birth
(YLD, years lived with disability)
Concept of YLL and YLD
Birth Death
(Life Expectancy At Birth = 80 years)
Life Expectancy At Birth
(YLL, Years of Life Lost = 30)
Life Expectancy At Birth
(YLD, years lived with disability = 15 years)
Nothing happens, guy lives for 80 years & dies
Guy dies @ age 50
Guy becomes *50% disabled* @ age 50
Concept of GBD and DALY
Why DALY
• Easy and comprehensive to conduct• Easy to understand, straightforward• Summary measure of population health• Led to increased attention to mental health• Led to increased attention to
noncommunicable diseases
Concept of DALY
DALY == DISABILITY/DEATH ADJUSTED LIFE YEARS
DALY = YLD + YLL
YEARS Lived with DISABILITYYEARS OF LIFE LOST DUE TO DEATHS
THIS CONCEPT IS APPLICABLE TO THE WHOLE POPULATION ACROSS AGE- AND SEX- AND FOR DIFFERENT DISEASE ENTITIES AND RISK FACTORS, AND DEATH
“Time" is the most appropriate metric
• Years Lived• Years lost due to death• Years lived in health states• Years lost due to health states
loss function
• Years lost as a function of the age at which death occured
• YLL(c,a,s) = N(c,a,s) * L(a,s)• N(c,a,s) = Number of deaths due to cause
“c”, given age “a” and gender “s”• Loss function is based on life expectancy at
age 80 (male), 82.5 (female)
Value choices (For Class Debate)
• How long should people in good health expected to “live”?
• National Level?• Globally?• Should there be time discounting?• 3% time discounting (in the last iteration,
removed in 2010 version)
Are lost years of healthy life valued more at some ages? (Class Debate)
• Real problems• Social Values and choices based on them• Age Weights• More weights for deaths at < 39 years
compared to olderyears• Most Contentious• These are removed in this iteration as well
YLL = Years of Life Lost
This is the simplest form of defining years of life lost from a population perspective.1. No of deaths at each age group (Age-specific death rate * Population)2. Standard Life Expectancy at that age of death
Comparison between Disease Ranking, using death rates versus YLL
Murray’s Value Choices: Questions
• How Long Should People in Good Health Expect to Live?– Will this be locally determined?– Will this be same for all people in the world?– ??
• How Should We Compare – Years of Life lost through death– Years lived with Disability? Poor Health?
YLL with Discount Rate Added
Years Lived with Disability
Disability Weight (DW) ranges between 0 and 10 == No Disability Whatsoever, full health1 == Completely Disabled, in other words, Dead (Death)•This is the simplest situation•Disability Weights are Societal Preferences
Disability Weights and Questions
• Disability Weights are Social Preferences, not values
• Is a Year of Healthy Life NOW worth more to Society than Healthy Life gained in FUTURE?
• Are Lost years of healthy life values MORE at some ages THAN at other ages? (What happens or should happen with old age and very young age?)
Example Weights
Measures of Health Gaps
• Lost years of full health• DALY is a health gap measure• permits categorical attribution
Concept of YLD (Years Lived with Disability)
• yld(c,a,s) = I(c,a,s) * dw(c,a,s) * l(c,a,s)• I(c,a,s) == age-sex stratified incidence• dw(c,a,s) == disability weight for c, for age a
and gender s• l(c,a,s) == duration in years of the disease c
until remission or death
Achievement of GBD and DALY Approach
• Quantified Burden of Disease• Developed Estimates of incidence,
prevalence, duration, case fatality• Charted over 500 sequelae• Analyzed physiological, social, behavioural
risk factors• Stratified by Age/Sex/Region
How is GBD 2010 Different Now from Earlier?
• DALYs in GBD 2010 has removed discounting and age weighting
• DALYs in GBD 2010 is simpler but more comprehensive
Calculation of DALYs in GBD 2010
• Has not included age weighting, no discounting
• Reference Standard population is based on lowest age-specific death rates across countries
• Disability Weights based on general public perception NOT experts
Philosophy of GBD 2010
• Do NOT allow advocates for the importance of specific diseases to choose the disability weights associated with specific disorders
• All sources of data have information• Internally consistent measures possible
Goal of GBD 2010
• Synthesise available data on the epidemiology of all major diseases+injuries
• Comprehensive and comparable assessment of the magnitude of 291 dieases+injuries
Methods and Study Design
• The world was divided into 21 epidemiological regions
• 20 age groups• Total of 291 diseases (diseases and injuries
== diseases)• Diseases are organized in four levels of
disaggregation
Methods (causes of death)
• YLL computed for 235 of 291 causes• YLL = N * L (Number of deaths * gap years
of life expectancy)• Two disorders only cause YLLs (SIDS, and
Ruptured Aneurysm*)
Sources of Cause of Death
• Vital Registration• Verbal Autopsy• Surveillance• Other sources (epidemiological studies,
surveys, etc)
How Death Rates for Various Diseases Were Calculated
• Used Cause of Death Ensemble Model (Consem model)
• Otherwise modeled deaths based on causal diagrams
• Obtained data from vital registration systems and other sources
• Caution: All cause mortality estimates must be consistent with the sum of cause specific mortality
• (Discussion Item: why is that important??)
Methods (YLD)
• YLD == Years Lived with Disability• Estimated for 1160 sequelae• YLD = Prevalence * Disability Weights
(Prevalence = Incidence * Duration in steady state)
Methods (Prevalence Estimation)
• Systematic Analysis of Published and available unpublished data sources
• Prevalence, Incidence, Remission, excess mortality
• Used DisMoDMR software to model the prevalence
Methods (Disability Weights)
• Measured for 220 unique health states• 1160 disease sequelae• Why are Number of Health States Lower than
Number of Sequelae?• Disability Weights were based on population
based surveys (N = 3100) worldwide• Disability Weights studies involved asking the
respondent to compare two alternative health states and rate them
How to Set up Rank Lists (the purpose of GBD)
• Choose the level in the cause hierarchy• They chose Second Level 21 causes• Another cause list of 176 causes• Causes were clustered under broader
categories• Regions were set up on the order of their
mean age of death (Why??)
So, what did they find?
• 2010: 2.49 Billion DALYs (360 dalys per 1000)• 31.2% from YLDs, 68.8% from YLL• YLD make little contribution in neonatal age
group• Global DALYs decreased slightly from 2·503
billion in 1990 to 2·490 billion in 2010• broad composition of the burden of disease has
shifted from communicable, maternal, neonatal and nutritional dis- orders to NCDs and injuries.
Top Ten Diseases, 2000 GBD Estimates
Top Ten Leading Causes of Disability, GBD 2001
Groups of Disease, GBD 2010
Country Wise Comparison of Distribution of Disease Impacts, GBD 2001
Comparison between profiles 1990-2010
GBD Comparison: 1990-2010
How Non-fatal Diseases are Gaining
Age Distribution of DALYs, GBD 2010
Implications
• Health system investment decisions• Capital investment of health system will need to
take into account the fact about shift from communicable to non-communicable disease load
• Content of Education?• Time to plan for better musculoskeletal disease
management (common, weighted higher, better assessed, new information uncovered)
Resources?http://www.healthmetricsandevaluation.org/tools/data-visualizations