Uncertainty

1
Uncertainty lessons from air pollution

• Jon Ayres
• Chair, ACP

2
Fog The London Incident 1952
4 000 excess deaths
3
Clean Air Act, 1956
4
The Six Cities Study
Risk of Death 1.1 10 increased
risk compared to Portage
5
Concerns..

• Essentially one study
• and that from the USA
• Effect sizes were small
• Although
• Good (we thought) exposure measures
• although not personal exposures
• And outcomes seemed clear (and logical)
• Pressure from NGOs (and others) to heed Tom
  Lehrer
• dont breathe the air!
• Or at least do something about it

6
Concerns..

• Essentially one study
• and that from the USA
• Effect sizes were small
• Although
• Good (we thought) exposure measures
• although not personal exposures
• And outcomes seemed clear (and logical)
• Pressure from NGOs (and others) to heed Tom
  Lehrer
• dont breathe the air!
• Or at least do something about it

Are these associations causal?
7
Bradford Hill and CausalityBradford Hill Proc
Roy Soc Med 196558295-300

• 1 Strength e.g. cigarette smokers have a 10 fold
  increased risk of lung cancer
• 2 Consistency i.e. same effects across different
  studies
• 3 Specificity i.e. disease virtually unique to
  exposure
• 4 Temporality i.e. the exposure must precede the
  effect
• 5 Biological gradient i.e. a dose response effect
  
• 6 Plausibility N.B. judged in the light of
  knowledge of the day
• 7 Coherence i.e. fits in with what is known
  already
• 8 Experiment i.e. a change in exposure leads to
  a change in disease
• 9 Analogy e.g. thalidomide other feto-toxic
  drugs

8
Bradford Hill and CausalityBradford Hill Proc
Roy Soc Med 196558295-300

• 1 Strength e.g. cigarette smokers have a 10 fold
  increased risk of lung cancer
• 2 Consistency i.e. same effects across different
  studies
• 3 Specificity i.e. disease virtually unique to
  exposure
• 4 Temporality i.e. the exposure must precede the
  effect
• 5 Biological gradient i.e. a dose response effect
  
• 6 Plausibility N.B. judged in the light of
  knowledge of the day
• 7 Coherence i.e. fits in with what is known
  already
• 8 Experiment i.e. a change in exposure leads to
  a change in disease
• 9 Analogy e.g. thalidomide other feto-toxic
  drugs

Confounders
9
Reducing uncertainty

• Population based studies
• Human exposure studies
• Animal studies
• Lab studies (cells largely)

10
UK Governments response

• To set up a committeeor three
• MAAPE - reports on episodes (1990)
• COMEAP - wider advisory remit (1991)
• EPAQS - AQ standards
  recommendations (1991)

11
Problems..differing opinion and inadequate data

• Variable methodology
• Personal exposures
• Imprecise metrics
• Confounding
• meteorology
• other pollutants
• co-exposures
• Thresholds
• Lags
• Transferability
• quantitative qualitative

• But outcomes more clear
• Deaths/HAs
• Heart and lung disease

12
Respiratory mortality and PM10 (n47)
13
Continuing uncertainties

• For instance
• Particles appear to be more toxic
• But have we got the right metric ?
• Gases might also act in concert
• Individual susceptibility
• Mechanisms
• Management
• Effects of long term exposures

14
Quantification of the health impact of air
pollution in the UK

• QUARK report (COMEAP 1998)
• GB urban
• Deaths brought forward (all cause)
• PM 8,100 pa
• SO2 3,500 pa
• O3 12,500 pa (no threshold)
• Respiratory hospital admissions brought forward
  additional
• PM 10,500 pa
• SO2 3,500 pa
• O3 9,900 pa (no threshold)

15
Estimates of the range of costs of air pollution
in the UK

• EAHEAP
• Total benefits (net of NHS costs of saved
  morbidity/mortality)
• PM10 0.93 - 540m pa
• SO2 0.45 - 440m pa
• O3 0.31 - 315m (summer only)
• DH, 1999

16
The time line of reducing uncertainty?
17
The time line of reducing uncertainty?





18
The time line of reducing uncertainty?

6 Cities study




19
The time line of reducing uncertainty?
Govt Cttees formed

6 Cities study




20
The time line of reducing uncertainty?
QUARK report
Govt Cttees formed

6 Cities study




21
The time line of reducing uncertainty?
EU stds
QUARK report
Govt Cttees formed

6 Cities study




22
The time line of reducing uncertainty?
EU stds
QUARK report
Govt Cttees formed

6 Cities study


O
O


x 1st EPAQS report (ozone) x last EPAQS
report (halides)
23
The time line of reducing uncertainty?
EU stds
QUARK report
Govt Cttees formed

6 Cities study



O
O


x 1st EPAQS report (ozone) x last EPAQS
report (halides)
24
Dealing with uncertainty lessons from air
pollution

• Work together
• multi-disciplinarity
• Assess the data for quality
• address causality
• where data are absent/inadequate - fill the gaps
• Try and quantify potential effects
• of both short- and long-term exposures
• Communicate