Air pollution and mortality:
Historical perspectives
Ross Anderson
St George’s, University of London
University of Hong Kong, April 2008
Outline
• Brief review of what we know about
outdoor air pollution and mortality and how
we got there
• Reflection on some of the most influential
conceptual developments
• Focus on concepts of causality and
mechanisms
2
Outdoor air pollutants
Gases Inhalable Particulates(PM10)
– Oxides of nitrogen – Fine (PM2.5) &Ultrafine (PM0.1)
– Ozone Created from Combustion
– Sulphur dioxide • Primary: carbonaceous
• Secondary: sulphates, nitrates
– Carbon monoxide
• Lead
“Air Toxics”
– Coarse (PM2.5-10)
– 1,3-butadiene
Mechanically generated
– Polycyclic organic matter e.g.
• Crustal source: silicaceous
PAH’s.
• Non combustion mobile sources:
– Benzene eg Tyre and brake wear
– Aldehydes, e.g. formaldehyde • Sea spray (coastal areas)
– Biological
• Pollen fragments, fungal spores
Outdoor air pollution and mortality
Short-term exposure (temporal) studies
• Episodes
– Increase in mortality
• Daily mortality studies
– Short-term associations with daily mortality
– Mainly cardiovascular and respiratory diagnoses affected
– Little specificity for specific diagnoses
– Not explained by short-term mortality displacement alone
– No threshold identified
– Cannot estimate life years lost
• Intervention studies
– Step reductions in air pollution are associated with short-term and long term
reductions in mortality.
• What source or component of pollution?
– Some evidence that particles are most important
– But generally low specificity for specific components or sources of pollution.
• Some coherence with temporal studies of morbidity and pathophysiological
effects
Outdoor air pollution and mortality
Long-term exposure (spatial) studies
• Spatial variations in chronic exposure to air pollution are
associated with increased risk of mortality even after
controlling for individual risk factors
– Cardiovascular and respiratory causes are both affected
– Emerging evidence of effects on infant mortality
– No threshold has been identified
– Relative risks of mortality are greater than for time-series studies
– Life years lost can be calculated
• What source or component of pollution?
– Some evidence that particles are most important
– But generally low specificity for specific components or sources
of pollution. 5
Air pollution and mortality:
Some uncertainties
• Roles of specific pollutants and sources
• Relationship between ambient concentrations and dose
to putative target organs
• Causality of epidemiological associations
• How air pollution actually increases mortality
6
Outdoor Air pollution and mortality
Evolution of epidemiology
• Pre 1850s
– Urban fogs associated with increased mortality
– Increased mortality in low lying urban areas attributed to miasma
• Pre 1930s
– Mortality in winter episodes attributed mainly to cold
• 1930/50s
– Formal investigation of major episodes.
• 1960s
– Spatial ecological studies
• 1970/80s
– Spatial studies with individual level confounder control
• 1980s
– Temporal studies of daily mortality
• 1990/2000s
– Intervention studies
– Impact assessment
– Effect modification 7
– Systematic review
Underlying trends affecting the development
of air pollution science
• Trends in methods and concepts
• Trends in the culture of science and the role of scientists
in society
• Trends in societal attitudes and priorities relating to the
environment and the causation of disease
• Trends in public administration
8
How does knowledge evolve?
Paradigm shifts
Thomas Kuhn (1962). The structure of scientific revolutions.
Origin of humankind: creation >>>> evolution
Physics: Newtonian >>>> Einsteinian
Pathophysiology: Galenic >>>> modern
Disease causation: miasma >>>> germ theory
Methods Concepts
Registration of deaths Causality
Environmental measurements and modelling Mechanisms
Biomedical measurements
Statistical theory
Data processing
9
Weighing the evidence of
observational studies
“Is there any other way of explaining the set
of facts before us, is there any other answer
equally, or more, likely than cause and
effect?” Bradford Hill (1965)
Temporality Size of effect
Dose response Specificity
Consistency
Coherence
Plausibility
Legal weight of evidence: Analogy
Balance of probabilities public health action
Beyond all reasonable doubt scientific acceptance
EVIDENCE ASSOCIATIONS:
FROM Air pollution and health
POPULATIONS
EXCLUDE:
Chance, Bias Confounding
OTHER
EVIDENCE APPLY:
(TOXICOLOGY) Scientific reasoning
JUDGEMENT:
Hill’s “viewpoints” Causality
“Precautionary principle” POLICY
Mechanistic theories
• Specific factors (e.g. Known carcinogens)
• Acute respiratory inflammation
• Secondary effects of respiratory
inflammation on cardiovascular system
• Chronic irritation
• Induction of asthmatic inflammation
• Systemic (i.e. beyond the respiratory tract)
• chemical mediator release
• neural stimulation
12
Mechanisms of PM effects
Pope and Dockery 2006
Mechanisms of PM effects
3
2.5
2 London 1952 smog: ratio
Ratio
1.5 of deaths in episode week
1
to previous week, by age
0.5
0
life years1 % reduction
20µg/m3
Baseline 26.7 0 11.5 0
Exposure 11.9 3.25m 17 £485m
reduction2
Exceedance 0 1.57m 13.4 Much higher
reduction
1 Cohort evidence: Pope et al. 6% hazard rate per 10µg/m3
2 Combined measures (EURO V, VI standards, penetration of low emission vehicles, small plant
combustion
The concept of multifactorial
causation
• Offers an explanation as to why small
exposures may have big effects
• Why do we not observe clear evidence of
a threshold in epidemiological studies
• Draws attention to the importance of other
causal factors increasing vulnerability
• Has profound implications for abatement
strategies
• Leads to difficulties in risk communication
31
Concluding remarks
• As in any other scientific field, knowledge of the
health effects of air pollution has increased as a
result of advances in methods and concepts.
• However, there are many things we are uncertain
about
– Pathophysiological and clinical mechanisms
– Relative toxicity of constituents
– The relationship between exposure measures and
actual dose to the target organ
– The possibility of residual confounding
• Do we need a paradigm shift? What could it be?
32
Thanks
33