Abstract

The NHS Abdominal Aortic Aneurysm (AAA) screening programme (NAAASP) was initiated in 2009 and invites men aged 65 in the UK to screening. The programme was initiated following the completion of the large Multicentre Aneurysm Screening Study randomised controlled trial and subsequent health economic modelling, which showed population screening for men to be highly cost-effective. AAA screening in women has, until now, not been considered worthwhile since the prevalence of AAA (aortic diameter ≥3.0cm) in women is substantially lower than in men. However, previous modelling has suggested that NAAASP would still be cost-effective in men even down to an AAA prevalence of 0.35%. Moreover, one-third of deaths from AAA in the UK are now in women. Hence the cost-effectiveness of AAA screening women needs to be formally assessed.

In this talk I will describe the Screening Women for Abdominal Aortic Aneurysm (SWAN) project, where we undertook a health economic assessment of AAA screening in women. A previous multi-state model of AAA screening in men was adapted to create a more flexible discrete event simulation (DES) model. Information from published literature or relevant databases was used to obtain input parameters for this model relevant to women.

I will describe how the DES model was developed to allow the progression of AAA to be modelled continuously using a mixed-effects growth model of aortic diameter. This allowed individuals to grow at different rates and for different surveillance, diagnosis and intervention strategies based on observed diameter measurements to be easily compared. Monte Carlo error in the estimation of incremental costs and effects was reduced using two complementary approaches; firstly we simulated pairs of “twins” with similar characteristics, one invited to screening and one not; and secondly we oversampled individuals who have an AAA at screening.

Finally, based on findings from this project I will address whether population AAA screening in women is likely to be cost-effective and provide recommendations for researchers wishing to conduct their own DES modelling.