Development and use of a Monte Carlo-Markov cycle tree model for coronary heart disease incidence-mortality and health service usage with explicit recognition of coronary artery revascularization procedures (CARPs) /

Abstract

[Truncated abstract] The main objective of this study was to develop and validate a demographic/epidemiologic Markov model for population modelling/forecasting of CARPs as well as CHD deaths and incidence in Western Australia using population, linked hospital morbidity and mortality data for WA over the period 1980 to 2000. A key feature of the model was the ability to count events as individuals moved from one state to another and an important aspect of model development and implementation was the method for estimation of model transition probabilities from available population data. The model was validated through comparison of model predictions with actual event numbers and through demonstration of its use in producing forecasts under standard extrapolation methods for transition probabilities as well as improving the forecasts by taking into account various possible changes to the management of CHD via surgical treatment changes. The final major objective was to demonstrate the use of model for performing sensitivity analysis of some scenarios. In particular, to explore the possible impact on future numbers of CARPs due to improvements in surgical procedures, particularly the introduction of drug eluting stents, and to explore the possible impact of change in trend of CHD incidence as might be caused by the obesity epidemic. ... When the effectiveness of PCI due to introduction of DES was increased by reducing Pr(CABG given PCI) and Pr(a repeat PCI), there was a small decline in the requirements for PCIs and the effect seemed to have a lag. Finally, in addition to these changes when other changes were incorporated which captured that a PCI was used more than a CABG due to a change in health policy after the introduction of DES, there was a small increase in the requirements for PCIs with a lag in the effect. Four incidence scenarios were developed for assessing the effect of change in secular trends of CHD incidence as might be caused by the obesity epidemic in such a way that they gradually represented an increasing effect of obesity epidemic (assuming that other risk factors changed favourably) on CHD incidence. The strategy adopted for developing the scenarios was that based on past trends the most dominant component of CHD incidence was first gradually altered and finally the remaining components were altered. iv The results showed that if the most dominant component of CHD incidence, eg, Pr(CHD - no history of CHD) levelled off and the trends in all other transition probabilities continued into future, then the projected numbers of CABGs and PCIs for 2001-2005 were insensitive to these changes. Even increasing this probability by as much as 20 percent did not alter the results much. These results implied that the short-term effect on projected numbers of CARPs caused by an increase in the most dominant component of CHD incidence, possibly due to an 'obesity epidemic' is small. In the final incidence scenario, two of the remaining CHD incidence components-Pr(CABG - no history of CHD) and Pr(CHD death - no CHD and no history of CHD) were projected to level off over 2001-2005 because these probabilities were declining over the baseline period of 1998-2000. The projected numbers of CABGs were more sensitive (compared to the previous scenarios) to these changes but PCIs were not.

Thesis (Ph.D.)--University of Western Australia, 2008

[Truncated abstract] The main objective of this study was to develop and validate a demographic/epidemiologic Markov model for population modelling/forecasting of CARPs as well as CHD deaths and incidence in Western Australia using population, linked hospital morbidity and mortality data for WA over the period 1980 to 2000. A key feature of the model was the ability to count events as individuals moved from one state to another and an important aspect of model development and implementation was the method for estimation of model transition probabilities from available population data. The model was validated through comparison of model predictions with actual event numbers and through demonstration of its use in producing forecasts under standard extrapolation methods for transition probabilities as well as improving the forecasts by taking into account various possible changes to the management of CHD via surgical treatment changes. The final major objective was to demonstrate the use of model for performing sensitivity analysis of some scenarios. In particular, to explore the possible impact on future numbers of CARPs due to improvements in surgical procedures, particularly the introduction of drug eluting stents, and to explore the possible impact of change in trend of CHD incidence as might be caused by the obesity epidemic. ... When the effectiveness of PCI due to introduction of DES was increased by reducing Pr(CABG given PCI) and Pr(a repeat PCI), there was a small decline in the requirements for PCIs and the effect seemed to have a lag. Finally, in addition to these changes when other changes were incorporated which captured that a PCI was used more than a CABG due to a change in health policy after the introduction of DES, there was a small increase in the requirements for PCIs with a lag in the effect. Four incidence scenarios were developed for assessing the effect of change in secular trends of CHD incidence as might be caused by the obesity epidemic in such a way that they gradually represented an increasing effect of obesity epidemic (assuming that other risk factors changed favourably) on CHD incidence. The strategy adopted for developing the scenarios was that based on past trends the most dominant component of CHD incidence was first gradually altered and finally the remaining components were altered. iv The results showed that if the most dominant component of CHD incidence, eg, Pr(CHD - no history of CHD) levelled off and the trends in all other transition probabilities continued into future, then the projected numbers of CABGs and PCIs for 2001-2005 were insensitive to these changes. Even increasing this probability by as much as 20 percent did not alter the results much. These results implied that the short-term effect on projected numbers of CARPs caused by an increase in the most dominant component of CHD incidence, possibly due to an 'obesity epidemic' is small. In the final incidence scenario, two of the remaining CHD incidence components-Pr(CABG - no history of CHD) and Pr(CHD death - no CHD and no history of CHD) were projected to level off over 2001-2005 because these probabilities were declining over the baseline period of 1998-2000. The projected numbers of CABGs were more sensitive (compared to the previous scenarios) to these changes but PCIs were not.