The Global Burden of The Global Burden of Disease Due to Urban Air Disease Due to Urban Air Pollution: Estimates and Pollution: Estimates and

UncertaintiesUncertainties

Aaron CohenAaron Cohen

Health Effects InstituteHealth Effects Institute

CRA project and WHR 2002CRA project and WHR 2002

www.who.int/whrwww.thelancet.com

Urban Air Pollution Working Urban Air Pollution Working GroupGroup

H Ross Anderson (Co-Chair) SGHMS U LondonAaron Cohen (Co-Chair) Health Effects

InstituteKersten Gutschmidt WHO/GenevaMichal Krzyzanowski WHO/Euro-BonnNino Künzli University of BaselBart Ostro California Health DeptKiran Dev Pandey World BankArden Pope Brigham Young Isabelle Romieu PAHOJonathan Samet Johns Hopkins Kirk Smith UC Berkeley

The Global Burden of Disease The Global Burden of Disease Due to Urban Air Pollution: Due to Urban Air Pollution:

Estimates and UncertaintiesEstimates and Uncertainties

• The Global Burden of Disease The Global Burden of Disease Comparative Risk AssessmentComparative Risk Assessment

• Estimating the Global Burden of Estimating the Global Burden of Disease Due to Urban Air PollutionDisease Due to Urban Air Pollution

• Magnitude and distribution of the Magnitude and distribution of the attributable burden attributable burden

• Major sources of uncertaintyMajor sources of uncertainty• Estimating the avoidable burdenEstimating the avoidable burden

Risk factors in CRAChild & maternal under-nutrition Childhood and maternal underweight Iron deficiency Vitamin A deficiency Zinc deficiency

Other nutrition-related risks & inactivity High blood pressure High cholesterol Overweight and obesityInadequate fruit and vegetable intake Physical inactivity

Addictive substances Smoking and oral tobacco Alcohol Illicit drugs

Sexual and reproductive health risks Unsafe sex Non-use and ineffective use of contraception

Environmental risks Unsafe water, sanitation, and hygiene Urban air pollution Indoor smoke from solid fuels Lead exposure Climate change

Occupational risks Risk factors for injury Carcinogens Airborne particulates Ergonomic stressors Noise

Other selected risks to health Contaminated health care injections Child sexual abuse

Distributions of risks by poverty

A: very low child and adult mortalityB: low child and adult mortalityC: low child, high adultD: high child, high adultE: high child, very high adult

14 WHO mortality sub-14 WHO mortality sub-regionsregions

Basic CRA framework and goals

Risk factor levels• current distribution• counterfactual

distribution(s)

Disease burden

Risk factor-disease relationships

• risk accumulation• risk reversal

Attributable burden in 2000Avoidable burden in 2010 & 2020

Mortality and Burden of Mortality and Burden of DiseaseDisease

• Mortality = Numbers of Deaths

• Burden = Disability Adjusted Life

Year or DALY

• DALY = YLL + YLD• years of life lost because of premature death

(YLLs)

• years of life lived with disability (YLDs)

• one DALY = one lost year of healthy life

Applying the CRA Methods to Applying the CRA Methods to Urban Air PollutionUrban Air Pollution

• Choice of indicator pollutant and estimation of ambient concentrations

• Choice of risk factor-disease relationships

• Calculation of disease burden

Estimated PMEstimated PM1010 Concentration in Concentration in World Cities (pop >=100,000)World Cities (pop >=100,000)

PM10 (µg/m3)

. 5-14

. 15-29

. 30-59

. 60-99

. 100-254

Population Distribution of Population Distribution of Estimated PMEstimated PM10 10 Levels for 3200 Levels for 3200

Cities Cities

Cohen et al 2004

American Cancer Society II Cohort American Cancer Society II Cohort 500, 000 adults followed 1982 – 500, 000 adults followed 1982 –

19981998 (Pope et al JAMA 2002)(Pope et al JAMA 2002)

Random effects Cox proportional hazards model controlling for age,sex, race, smoking, education, marital status, body mass, alcohol,occupational exposure and diet

RR per 10µg/m3 PM2.5 1979-83

RR 95% CI

Cardiopulmonary

1.06 1.02-1.10

Lung Cancer 1.08 1.01-1.16

Alternative Scenarios for Burden of Alternative Scenarios for Burden of Disease Estimation for Urban Air Disease Estimation for Urban Air

PollutionPollution

Alternative concentration-response curves for cardiopulmonary deaths

PM2.5 (g/m3)

0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90

RR

1.00

1.25

1.50

Base Case, PM2.5 Max=50

PM2.5 Max=30Linear ExtrapolationLog-linear Extrapolation

Counterfactuallevel of 7.5 g/m3

Percent change in mean daily Percent change in mean daily number of child and infant deathsnumber of child and infant deaths

-4.0

-2.0

0.0

2.0

4.0

6.0

8.0

Sao Paulo

Mexico City

Sao Paulo

Sao Paulo

Sao Paulo

Bangkok

Fixed-Effe

cts

Random-Effe

cts

% c

han

ge

for

10 u

nit

in

crea

se (

ug

/m3)

in

PM

10

1. Calculate region specific relative riskRR2.5 = exp [CR * (X – 7.5)]where CR is slope of the C-R function (β

coefficient) and X is regional population weighted mean PM.

2. Calculate Attributable Fraction (AF)

AF = P(RR-1) / [ P(RR-1)+1]where P is proportion exposed, i.e.

proportion living in cities3. Calculate attributable deaths and DALYs(AF * regional totals)

Estimation of attributable deaths Estimation of attributable deaths and DALYsand DALYs

Estimated Burden of Urban Air Estimated Burden of Urban Air Pollution Worldwide Pollution Worldwide

(95% confidence intervals)(95% confidence intervals)

AF(%)

Deaths(thousands)

YLL(thousands)

CPD 4(1, 6)

712(245, 1107)

4,666(1,695, 7700)

Lung Ca 5(1, 9)

62(10, 114)

572(92, 1,063)

ARI (< 5yr.) 1(-1, 3)

26(-24, 66)

862(-799, 2,228)

All-cause(from time-series studies)

1(NA)

378(NA)

(NA)

Fraction of Deaths Attributable to Fraction of Deaths Attributable to Outdoor Urban Air Pollution by RegionOutdoor Urban Air Pollution by Region

0123456789

10

%

LCA CPD ARI<5 yrs

Mortality attributable to leading risk factors

0 1000 2000 3000 4000 5000 6000 7000 8000

High blood pressure

Tobacco

High cholesterol

Underweight

Unsafe sex

Low fruit and vegetable intake

Overweight and obesity

Physical inactivity

Alcohol

Unsafe water, sanitation, and hygiene

Indoor smoke from solid fuels

Iron deficiency

Urban air pollution

Zinc deficiency

Vitamin A deficiency

Contaminated health care injections

Occupational airborne particulates

Occupational risk factors for injury

Lead exposure

Illicit drugs

Mortality in thousands (Total 55.86 million)

High-mortality developing

Lower-mortality developing

Developed

Ezzati et al. 2002; WHO 2002

Summary results for individual Summary results for individual risksrisks

• Substantial disease burden associated with risk factors such as under-nutrition, poor water and sanitation, and indoor air pollution remain, especially in the poorest developing countries

• Simultaneously risks from a number of factors such as smoking, alcohol, and obesity are becoming increasingly global

• Some risks, like urban air pollution and lack of contraception are major causes of burden in specific regions

Excess Deaths from Selected Excess Deaths from Selected Environmental FactorsEnvironmental Factors

Uncertainties that we quantifiedUncertainties that we quantified

• Random variation in exposure and risk coefficient estimates

• Choice of PM2.5/PM10

• Choice of counterfactual level

• Choice of concentration-response function: coefficients and extrapolation

Sensitivity of Attributable Sensitivity of Attributable

Mortality EstimatesMortality Estimates

-40 -20 0 20 40 60 80

% relative to base case

Trunc at 30

Full linear

Log-linear

Base 2.5/10

Min 3

Min 15

Av ACS coeff

CP LungCa

Uncertainties that we did not Uncertainties that we did not quantifyquantify

• Burden due to pollutants other than PM, e.g., ozone

• Burden in cities with populations <100 K

• Effects of exposure at finer spatial scales e.g., due to proximity to vehicular traffic

• Relative toxicity of PM from different sources

• Contribution of other potentially important health outcomes, e.g., LBW, infectious disease

Leading Causes of Mortality and Burden of Disease 2002

%• Ischaemic heart disease 12.6• Cerebrovascular disease 9.7• Lower respiratory infections 6.8• HIV/AIDS 4.9• COPD

4.8• Perinatal conditions 4.3• Diarrhoeal diseases 3.2• Tuberculosis 2.7• Lung cancer 2.2• Malaria 2.2• Road traffic accidents 2.1

%• Perinatal conditions 6.5• Lower respiratory infections 6.1• HIV/AIDS 5.7• Depression 4.5• Diarrhoeal diseases 4.2• Ischaemic heart disease 3.9• Cerebrovascular disease 3.3• Malaria 3.1• Road traffic accidents 2.6• Tuberculosis 2.3• COPD 1.9

Mortality DALYs

Pediatrics 2005;115:121-128

Tuberculosis All Forms, WHO Regions, 2002

Africa27%

the Americas4%

Eastern Mediterranean

7%

Europe5%

South-East Asia33%

Western Pacific24%

Tuberculosis Deaths in HIV-Negatives, 2002

Africa19%

the Americas3%

Eastern Mediterranean

8%

Europe5%South-East

Asia43%

Western Pacific22%

1.6 Million Deaths

8.9 Million Cases

Air Pollution and TBAir Pollution and TB

Modified from Baris and Ezzati 2004

Millions of Deaths(% in

developing countries)

% Global Burden of Disease

(% in developing countries)

Range of reported TB relative riskestimates

Tuberculosis 1.6 (>90%)

2.5(>90%)

Smoking 4.8(50%)

4.1(>50%)

1.5-4.5(ever/never)

Indoor Air Pollution

1.6(>95%)

2.6(>95%)

1.8-3.6(solid fuel

use)

Urban Outdoor Air Pollution

0.8(>70%)

0.4(>70%)

None reported

c

d

Unavoidable

Dise

ase

Burd

en

Time

T0 TxPast Future

0%

25%

50%

75%

100%(Theoreticalminimum)

a

b

Exposure reductionat T 0

Attributable and avoidable Attributable and avoidable burdenburden

What determines the health What determines the health effects of air pollution as effects of air pollution as

economies grow?economies grow?

Number of people in cities

Sources of air pollution

Emissions

Air quality

Susceptibility

A: very low child and adult mortalityB: low child and adult mortalityC: low child, high adultD: high child, high adultE: high child, very high adult

Number of people at high CV risk 2000 - 2010Number of people at high CV risk 2000 - 2010(A Rogers 2005)(A Rogers 2005)

5 m

4 m

38 m

19 m 1 m

5 m 12 m

46 m

18 m

36 m

7 m

45 m16 m

50 m

>175 million people at 25%+ risk of a

major CV event in the next decade, by WHO

subregion

Reductions in Mortality in Dublin 72 Reductions in Mortality in Dublin 72 Months Pre- vs. Post-1990 Ban on Coal Months Pre- vs. Post-1990 Ban on Coal

SalesSales

Clancy et al. Lancet 2002Clancy et al. Lancet 2002

-20

-15

-10

-5

0

5

Adj

uste

d %

cha

nge

Total non traumaCardiovascularRespiratoryOther

Thank You

acohen@healtheffects.org