Norwegian Cardiovascular Disease Model (NorCaD) – a simulation model for estimating health benefits and cost consequences of cardiovascular interventions
Health technology assessment
|Updated
Key message
This report provides an introduction to the NorCaD cardiovascular model. The model was constructed for use in health technology assessments (HTAs) of prevention strategies directed towards cardiovascular disease. NorCaD is a state transition model which follows individuals from before they have any symptoms of cardiovascular disease until death.
• The model is based on Norwegian data on incidence of primary cardiovascular events and adapted to a Norwegian health care setting.
• Probabilities of disease progression are to a large extent based on data from international registries and randomised trials.
• Unit costs are gathered from Norwegian official data and resource use is mainly based on expert opinion.
• The model has a cycle length of one year, and runs from age 30 to 100 or to death.
As the cardiovascular disease progresses, costs, quality of life and life years are recorded to give the opportunity of cost-effectiveness-analyses.
The model is validated to fit Norwegian mortality data as close as possible. However there are still limitations regarding the model.
Summary
Background
Cardiovascular disease (CVD) is the most frequent cause of death in all developed countries and most other as well. In Norway, about 40% of all deaths are attributed to CVD and the population life expectancy would increase by about 4 years if all CVD were eliminated. For decade’s doctors, researchers, pharmaceutical companies, governments and others have struggled to reduce the burden of CVD. A range of new interventions has been proposed, and several are in use. The development of new interventions continues, but not all improvements have a substantially increased effect compared with older treatments, and some are costly. In Norway, the Patients’ Right Act grants patients the right to treatment, but only if the costs are reasonable in relation to the health benefits. It is therefore a need to quantify costs and benefits of CVD interventions.
The objective of the current project was to develop a model of atherosclerotic CVD from its asymptomatic stage through various CVD events and complications to death, and capture data on life years, quality of life and costs.
Methods
We used the software program TreeAge Pro to develop a transition model (Markov model) with cycles of one year from the age of 30 years to death or the age of 100. The model starts with all individuals free from symptoms of cardiovascular disease. All individuals are at risk of having one or more of the following primary CVD events: acute myocardial infarction (AMI), stroke, angina pectoris or heart failure. The risks of these events are as far as possible based on population data from Norwegian registries. After an event, patients move to one of the following health states: asymptomatic CVD (post CVD), heart failure, stroke sequelae and death. While patients are in any of these states (except dead), they are at risk of secondary CVD events.
The risks of secondary cardiovascular events are based on data from both registries and a range of randomised trials. Because randomised trials usually include patients with less severe health profiles, we adjusted data from these randomised trials to better represent average patients.
A Markov model has no built-in memory of previous disease-history. In some health states, however, it is reasonable to assume that previous disease history affects the probability of new events. To overcome this problem, we divided some health states (asymptomatic, stroke sequelae and heart failure) into more health states to capture differential risks. We also divided the health state dead into death from CVD and death from other causes to account for causes of death.
The effect of interventions increases life expectancy by the reduced risk of cardiovascular events and death. In the model these effects are relative risk reductions which are based on systematic reviews of interventional effects.
All costs related to CVD events or states are recorded as the disease progresses. Data on unit costs were taken from official Norwegian sources where possible. Data on resource use was to a large extent based on expert opinion, and to some extent on published data.
The model measures health outcome as life years based on the mortality risks built into it. To the extent decision makers request data on quality adjusted life years, the model allows for such outcome by assigning quality weights to the different health states.
Validation
We validated the NorCaD model by fitting model survival to survival in the Norwegian population. After validation, the model gave expected remaining lifetime less than 1% away from predictions by Statistics Norway.
Discussion
The NorCaD model was designed for economic evaluation of primary CVD prevention, but can also be used for secondary prevention. The model is comprehensive in terms of potential events and health states, but is created for a Norwegian setting. The strength of the model is its complexity and ability to analyse a wide range of interventions. Such abilities, however, require a wide range of input data. In total the model has about 200 parameters each with its own uncertainty. It should be noted, however, that any decision on CVD intervention is implicitly based on such uncertain information. By modelling and quantifying costs and out-comes, the uncertainties become explicit.
The model is based on recent Norwegian incidence data, which is an advantage compared to the conventional use of Framingham data that are older and taken from another country (USA).
Conclusion
The NorCaD model is a comprehensive and validated decision-analytic model which has potential to be used in several settings related to cardiovascular disease in Norway and else-where.