Modelling burdens attributable to specific diseases and causes

1
Modelling burdens attributable to specific
diseases and causes

• John Powles
• Jwp11_at_cam.ac.uk

2
Nature of causation in epidemiology
most common
3
Necessary, sufficient and component causes of
disease (Rothman)
A is a necessary cause
4
Population attributable fraction (risk) or burden

• For a dichotomous (harmful) exposure
• Proportion that would not have occurred with zero
  exposure
• But
• Needs also to be generalised to continuous
  exposures (eg adiposity, bp)
• And
• To preventive exposures eg Physical activity

5
Population attributable fraction (risk) or burden

• Generalising to continuous exposures
• attributable burden
• difference between burden currently observed and
  what would have been observed under a (past)
  counterfactual exposure distribution

6
Generalising to preventive exposures

• For a dichotomous protective exposure
• Proportion of the cases that would have occurred
  in the absence of exposure that were prevented by
  the exposure
• Note denominator is the hypothetical total
  applying in the unprotected counterfactual
• EG for moderate alcohol drinking and IHD
• Prevented fraction Prevented cases
  /Total expected in counter-
  factual non-drinking
  population

7
Generalising to preventive exposures

• Generalising to continuous exposures
• prevented burden
• difference between burden currently observed and
  what would have been observed under a (past)
  counterfactual exposure distribution

8
Avoidable burden

• More relevant to current policy are the potential
  future benefits of deciding to do something now

9
Avoidable burden

• Reduction in future burden expected if current
  exposure changed to a specified (more favourable)
  counterfactual exposure

10
Counterfactual exposure distributions

• Theoretical minimum risk
• Plausible minimum risk
• Feasible minimum risk
• Cost-effective minimum risk

11
(No Transcript)
12
Counterfactual exposure distributions

• Theoretical minimum risk
• Eg Tobacco zero smoking
• Alcohol 1-2 drinks/d for 100 of middle-
  aged and
  older
• ? Such distributions often implausible for
  public health purposes

13
Counterfactual exposure distributions

• Plausible minimum risk
• Ie Possible to imagine a society with such a
  distribution of risk
• Eg Tobacco smoking limited to a very small
  minority
• Alcohol a distribution with (say) 70
  drinking up to 1.5 d/d 15 abstainers

14
Counterfactual exposure distributions

• Feasible minimum risk
• Ie a distribution of risk that has been actually
  been attained by some society
• Eg Tobacco (once adopted) 80 non-smokers
• Alcohol a distribution with 55
  drinking up to 1.5 d/d 20 abstainers

15
Counterfactural exposure distributions feasible
minimum risk
16
Counterfactual exposure distributions

• Cost-effective minimum risk
• Ie A distribution of risk after a specified
  economically and politically feasible control
  programme
• Eg Tobacco and alcohol based on overviews of
  effectiveness of policies including tax and
  legislative measures

17
Attributable burdens are unbounded (ie not
constrained to add to 100)

• Eg Asbestos exposure smoking as causes of lung
  cancer in asbestos workers

From Hammond et al, 1979
18
Implications of unbounded nature of the sum of
attributable/avoidable burdens

• Plausibility is only constraint
• All such claims should be interpreted
  appropriately and cautiously

19
Attributable burden for a determinant

• AB S AFjBj
• Where
• AFj is the attributable fraction for condition j
  from the determinant of interest and Bj is the
  burden from condition j

20
The potential impact fraction (PIF) for a
stratified exposure

• For m exposure strata, where
• P and P are proportions in a given exposure
  stratum actually and counterfactually
• R is the RR for the given exposure stratum

Kleinbaum et al, 1982
21
Scan of murray and lopez 1999 p 599
22
For scan of murray and lopez 1999 p 596