Global burden of disease Putting NCDs in the

Global burden of disease Putting NCDs in the health agenda in LIC/MIC Pascal Bovet, MD, MPH University Institute of Social & Preventive Medicine, Lausanne Consultant for NCD, Ministry of Health, Seychelles WHO-IUMSP International Seminar on the Public Health Aspects of NCDs Lausanne, 10 -18 August 2010

Outline • • • The Global Burden Disease Project DALYs Sources of data Coherence in putting epidemiological GBD data together Assumptions underlying DALYs estimates Limitations Results on burden of disease related to diseases Results on burden of disease related to risks Ongoing review of GBD Conclusions

What is burden of diseases analysis (BOD Project)? • Collaboration between WHO, World Bank and Harvard SPH • A standardized framework integrating all available epidemiological information on morbidity and mortality • All morbidity and mortality estimates are made internally consistent (e. g. all deaths add to 100%) using conditionspecific epidemiology • Provides a quantitative assessment of the distribution of both diseases and their causes (RF) in populations • Common single metric to quantify health status (DALY) Comparable estimates across causes Consistent measure for use with intervention analyses • Allows to assess to which extent disease and risk for diseases could be avoided in different populations/regions

Disability-adjusted life years (DALY) DALY= YLL + YLD Years of Lost Life (due to premature mortality) Years Lost to Disability (due to injury or illness) One DALY = one lost year of healthy life • DALY combines information on mortality and morbidity in a single number (time as common metric for mortality and health states) • Assumes same optimal life expectancy throughout the world and that death/disability is only due to age and sex -not income, location or social class- ie everyone has the right to optimal health) • Several assumptions (see below)

n t r o d u c t i o n 1 Example of simplified DALY calculation (not taking into account age disease weighing, time discount) Example A: • 100, 000 children are stricken for 1 week with a disease with a disability weighting* of 0. 3; 2% die at 1 year old. • DALYs = YLL + YLD = (2000 X 80) + (100, 000 X 7/365 X 0. 3) = 160, 000 + 575 = 160, 600 Example B: • 100, 000 adults are stricken for 2 years with a disease with disability weighting* of 0. 6; 20% die at age 80 years. • DALYs = YLL + YLD = (20, 000 x 2) + (100, 000 x 2 x 0. 6) = 40, 000 + 120, 000 = 160, 000

Sources for GBD estimates of mortality and causes of death • Mortality level – Vital registration and census – Surveys – Demographic models • Causes of death – Vital registration with certification of cause of death – Sample registration systems – Verbal autopsies – Household surveys – Population surveillance systems

Numbers of datasets in 2001 update of the GBD Death registration data Child/adult Epidemiologic mortality data sources Total data sources Asia/Pacific 117 118 1, 820 2, 055 Europe 149 22 971 1, 142 High income Latin America & Caribbean Middle East & North Africa Sub-Saharan Africa 142 16 1, 830 1, 988 286 122 1, 311 1, 719 46 67 645 758 30 190 2, 185 2, 405 World 770 535 8, 747 10, 052 Ezzati (personal communication)

Availability of vital registration data Few data in several regions limiting inference

Quality of cause-of-death from national registration, 2003 Few data in several regions limiting inference

GBD challenge: assessing a definite number of conditions, their epidemiology and their disability • Mortality and morbidity for 107 conditions • identify age of occurrence, duration, incidence, prevalence, case fatality, death rate for each complications by sex, age and region • 483 complications for 107 conditions by sex, age & region: • e. g. 5 for DM: DM per se, retinopathy, neuropathy, diabetic foot, amputation • Estimation of disability of complications with weighing factor between 0 (perfect health) et 1 (same severity as death) • Extreme complexity underlies potential but also limitations (data in LIC) Murray & Peto, Lancet 1997

Data needed to estimate Burden of Diseases Þ Mortality = Years of life lost (YLL) vs optimal age of death: mortality and age at death by age, sex, & region Þ Morbidity = Years lived with disability (YLD) vs optimal age of death: incidence, remission, duration, sequelae, associated disability, severity, case fatality by age, sex and region Þ Disability adjusted life years (DALY): YLL + YLD

BOD: Coherent estimates linking incidence, case-fatality, remission, and prevalence for 107 conditions (Dis. Mod) Susceptible for case disease S(t) i m Death from other cause m d. C(t) = i. S(t)-(r+f+m)C(t) d. M(t)/dt = m(S(t)+C(t) d. D(t)//dt = (C)t r Case with specific disease (t) d. S(t)/dt = -(I+m)S(t) (by sex, age and region) f Death from case disease The Global Burden of Disease, WHO, Harvard, World Bank, 1996 Develop coherent estimates over 40 person years Two 1000 -page books with such estimates

Assessment of disability • Before BOD, no method to assess disability in different diseases • Weight from 0 (disability ~death) and 1 (perfect health) • Different ‘trade-off ’ methods (expert panels, e. g. with analog scales, ranks, time trade-off , etc) • Large consensus between 8 panels in different panels (r >0. 9) Lancet, 1997; 349: 1350

Methods for assessing disability • Ratings scales: visual analogue (place various health states on line with 0=equivalent to death and 1=perfect health) • Standard gambles: offer 2 alternatives: normal health for x years vs. poor health for condition C for y years (cave: possible bias by gamble like/dislike but others find it the reference as it is based on von Neuman-Morgenstern axioms of expected utility) • Time-trade off: 2 choices and alter time until one is indifferent between 2 choices • Person trade-off: would a decision-maker prefer to save x people with health state a vs. y people with state b (cave: sometimes problem for persons to make choices for others) • Issues: types of respondents (health prof, public, decision makers); adaptation (function), coping (changing expectations & norms) & adjustment (shift of valuation to those possible)

Seven disability categories Lancet, 1997; 349: 1350

Duration of YLD and YLL is scaled to standard expected years of life lost (LE 82 W, 80 M) • SEYLL = life expectancy at each age is a standard estimated duration of life expected at each age • Standard rates are from Coale and Demeny Model level 26 (used in BOD) World Health Report 1997; GBD study, WHO & WB, 1996. Mc. Kenna MT et al U. S. Burden of Disease–Past, Present and Future. Ann Epidemiol 2009; 19: 212– 219

Age weighting: value of a year of life is different at different ages in the Global Burden of Diseases Age weighting: C*x*exp(- x) with =0. 04; C=0. 1658 World Health Report 1997; GBD study, WHO & WB, 1996

Discount rate on future life largely underlie potential benefit of future interventions (BOD = 3%)

Calculation of DALYs in BOD DALYs = YLL (years of life lost) + YLD (years lived with disability) Parameters: a= age at death; r = discount rate; b = parameter for age weighting; C =: constant; L= standard expectation of life at age a. For GBD: r=0. 03; b=0. 04; C=0. 166; K=1. Life expectancy=82. 5 y (W), 80 y (M) Murray. The Global Burden of Disease. WHO, Harvard, WB, 1996

GBD: modeling trends in mortality and disability • 4 distal determinants explain large variance in age-, sex-, and cause-specific mortality: income per capita, human capital (education), smoking intensity and time • Regression equations for GBD estimated for 98 detailed causes from registration data from 67 countries • ln. Ma, k, i = Ca, k, i + 1 ln. Y 2 ln. HC + 3 T • M= mortality for age group a, sex k and cause i; Y= GDP per capita; HC=human capital (y of school); T= time World Health Report 1997; GBD study, WHO & WB, 1996

Relating disease burden to risks (vs. diseases) • An alternative way of looking at ill health is to consider risks for disease rather than diseases themselves • A single risk factor may underlie several diseases; e. g. unsafe sex for HIV, other STDs, and teenage pregnancy • Information on risk is most useful to direct health interventions • “Risk factor levels of today are disease or tomorrow” • DALYs have been assigned to health risks as well as to diseases

The Comparative Risk Assessment (CRA) Project • Quantified the role of selected risk factors in global and regional burden of disease using comparable methods • Coordinated at WHO and Harvard 1999 -2004

Risk factors in Comparative Risk Assessment Project Child & maternal under-nutrition Child and maternal underweight Iron deficiency anaemia Vitamin A deficiency Zinc deficiency Suboptimal breastfeeding * Term IUGR * Sexual and reproductive health risks Unsafe sex Non-use and ineffective use of contraception (unwanted pregnancy) Environmental risks Unsafe water, sanitation, and hygiene Urban air pollution Other nutrition-related risks & inactivity Indoor smoke from household solid fuel use High blood pressure Lead exposure High cholesterol Global climate change High blood glucose * Occupational risks Overweight and obesity Risk factors for injury Inadequate fruit and vegetable intake Carcinogens Physical inactivity Airborne particulates Addictive substances Ergonomic stressors Smoking and oral tobacco use Noise Alcohol use Other selected risks to health Illicit drug use Contaminated health care injections * Subsequent analysis Child sexual abuse

Issues with calculation of GBD • Limited quality of data in several regions (e. g. mortality in LIC) • Issues on assumptions (age weights, disability, discount, LE) • Projections based on 4 variables (time-demography, income, education, smoking): but should other variables be considered? Has been instrumental to put NCD in agenda

Findings in the BOD: Mortality and morbidity can be attributed to both diseases and risk factors 1. 2 million lung cancer deaths (of 56 million global deaths) Occupational exposures 10% Smoking 71% Air pollution 7% Poverty or education Ezzati

Burden of diseases worldwide in 2005: Large proportion related to NCD • Total: 52 million • NCD: 35 million (60%) • • CVD : 17. 5 million Cancer: 7. 5 million Chronic respiratory disease: 7. 5 million Diabetes: 1. 1 million • HIV: 3 million • TB: 1. 6 million • Malaria: 0. 8 million WHO Global Report "Preventing chronic diseases: a vital investment". WHO, Geneva, 2005 http: //www. who. int/chp/chronic_disease_report/en NCD Infectious diseases

Disease burden (DALYs) per 1000 pop. by region, 2001 Burden dominated by infectious diseases in LIC (young age distribution!) Lopez et al Lancet 2006

Probability of broad-cause death by age and region (men) High age-specific NCD death rates in adults in all regions Group 1: infectious diseases; Group 2: NCDs; Group 3: injuries Lancet, 1997; 349: 1269

Broad-cause death rates by income group at age 15 -59: Higher death rates of NCD in adults in LIC than HIC Launched 27 October 2008

Magnitude of premature (<70 y) NCD deaths by income group Large number of “preventable” NCD deaths in MIC/LIC

Broad-cause death trends by income group, 2004 -2030: Large and increasing proportions of NCD deaths in all income groups Global Burden of Disease 2004 Update: WHO 2008 (http: //www. who. int/healthinfo/global_burden_disease/2004_report_update/en/)

Deaths attributable to risk factors by income groups NCD risk factors account for large number of deaths in all income groups World Health Report, WHO, 2002

Ranking of 10 leading risks of death by income group, 2004 NCD risk factors also contribute high death rates in LIC (Figures should also be considered when adjusting for age) Global health risks: mortality and burden of disease attributable to selected major risks. WHO 2009

Global mortality and burden of disease attributable to CVD and their major risk factors, age 30+ (100 countries) A few NCD risk factors account for large BOD in adults Ezzati M et al Rethinking the ‘‘Diseases of Affluence’’ Paradigm: Global Patterns of Nutritional Risks in Relation to Economic Development. PLo. S 2006

Global Burden of Diseases and Risk Factors 2005 • Objective 1: Estimates of burden of diseases, injuries, and risk factors for 1990 and 2005 within 21 regions using consistent and comparable methods, assumptions, and data sources • Objective 2: Create simplified analytical tools to facilitate national burden estimates and policy use, and create curricula and training tools to teach the application and interpretation of GBD • Launched in September 2007 with an announcement in the Lancet • 40 -50 expert working groups for different diseases, injuries, and RF • Attempt to have comprehensive set of risks for major disease groups (CVD, selected cancers, major childhood diseases, etc) • Results to be published in 2010 and subsequently Ezzati, personal communication

Conclusions • BOD is an innovative method to assess burden of disease worldwide in a comprehensive, comparable and consistent manner • As LE rises, mortality rates alone insufficient, particularly for NCD, need to include non-fatal health outcomes in the policy debates • Single metric integrating mortality & morbidity allows direct comparison of global BOD (priority areas, guide cost-effective interventions, etc) • Several challenges: lack of data in some regions, assumptions (DALYs) • Continuous attempt to improve methods and results (update in 2010) • Double burden still an issue in many countries • BOD put NCD on agenda as NCD rank high in all regions • BOD (DALYs) shows high burden of preventable NCD (mental, dementia) and further prompts prevention of NCD that can be prevented • Links of BOD to both diseases (“risk factors of yesterday”) and risk factors (“disease of tomorrow”) critical to guide policy and interventions

Age-standardized NCD burden by income group, 2004 Large share of neuro-psychiatric conditions in BOD (when considering DALYs)

Deaths and DALYs attributable to 6 risk factors, by income region, 2004 Substantial prop. of deaths are related to NCD RF in LIC (Need to adjust for age of populations? ) Global health risks: mortality and burden of disease attributable to selected major risks. WHO 2009