Table 1 Sources of data for model parameter estimates
ParameterSource
Lung cancer
Incidence probability
    SmokersAustralian national incidence data for 2001 from the National Cancer Statistics Clearing House,34 adjusted on the basis of the RR of lung cancer for smokers versus never-smokers,35 and the estimated historical prevalence that gave rise to the current lung cancer epidemiology.7 35
    QuittersSmokers’ probabilities reduced over time after quitting according to an exponential model.7
Probability of deathVictorian Cancer Registry data for 1994–9.7
Utility of lifeThe mean of two estimates from the Harvard Catalogue of preference scores,11 and 2 more recent studies.36 37
Healthcare costsEstimated from an Australian study,38 and a more recent study from the United Kingdom.7 39
AMI
Incidence probability
    SmokersPopulation incidence estimated from Australian national mortality and hospitalisation data for 2001,40 41 then adjusted on the basis of the RR of AMI for smokers versus never-smokers,42 and the prevalence of smoking in 2001.7 43
    QuittersSmokers’ probabilities reduced over time after quitting according to an exponential model.7 28
Probability of deathEstimated from population mortality rates,40 and incidence probabilities (above) using DisMod II.7 44
Utility of lifeThe mean of seven estimates from the Harvard Catalogue of preference scores.7 11
Healthcare costsEstimated from hospitalisation costs,28 and assumptions about medications and ambulatory care post-discharge.7
Stroke
Incidence probability
    SmokersNEMESIS, an Australian population-based study of stroke incidence,45 adjusted on the basis of the RR of stroke for smokers versus never-smokers,42 and the prevalence of smoking in 2001.7 43
    QuittersSmokers’ probabilities reduced over time after quitting according to an exponential model.7 28
Probability of deathPerth Community Stroke Study.7 46 47
Utility of lifeA meta-regression of 20 articles.7 10
Healthcare costsNEMESIS, an Australian population-based study of stroke,48 adjusted to 2001 dollars.7 15
COPD
Incidence probability
    SmokersAustralian incidence for 2001 estimated from COPD prevalence data using DisMod II,7 then adjusted on the basis of the RR of COPD for smokers versus never-smokers,42 and the estimated historical prevalence of smoking, as for lung cancer.7 35
    QuittersSmokers’ probabilities reduced over time after quitting according to an exponential model.7
Probability of deathEstimated from population mortality rates,40 and incidence probabilities using DisMod II.7 44
Utility of lifeThe mean of three estimates from the Harvard Catalogue of preference scores.7 11
Healthcare costsThe Canadian Confronting COPD study,9 adjusted to Australian dollars on the basis of Purchasing Power Parities.49
Probability of death from other causes
    SmokersAustralian mortality data,40 adjusted on the basis of the RR of mortality for smokers,50 and the prevalence of smoking in 2001.7 43
    QuittersSmokers’ probabilities reduced over time after quitting according to an exponential model.7
  • AMI, acute myocardial infarction; COPD, chronic obstructive pulmonary disease; RR: relative risk.