Title: A microsimulation model of the BreastScreen Australia program
Abstract: The BreastScreen Australia breast cancer screening program uses biennial mammography (breast x-rays) to screen for breast cancer in the non-symptomatic population. Breast density is a measure of how much of the mammogram is occupied by usually normal breast tissue that is distinctly radiodense. The performance of the screening program is currently inferior for women with high breast density: they are found to have larger tumours at detection and higher rates of cancers detected between scheduled screens. More frequent screening for women with higher breast density might bring some benefit. Breast density varies substantially in the population and within women over time, and high breast density is associated with breast cancer risk. Given the complex associations between breast density, cancer risk, and individual-level factors such as age, menopause and hormone therapy use, microsimulation modelling is a useful tool to examine the potential benefit of alternative screening strategies for women with high breast density. We specify a continuous-time, stochastic, single-cohort microsimulation model of the BreastScreen Australia program, which we name the Australian Breast Cancer Screening Simulation (ABCSS). ABCSS is driven by observed data from over 10,000 BreastScreen Australia participants. ABCSS is specified so that it predicts detection rates and tumour size at detection, in women who develop breast cancer and participate in screening. It includes an individual-level model for breast density according to age, menopause and hormone therapy use, and a sub-model for the probability of cancer detection as a function of tumour size and breast density. We use ABCSS to simulate trials of screening strategies, where more frequent screening is offered to women with breast density above a specified threshold, and this is offset by less frequent screening for women with breast density below a specified threshold such that the number of screening services required does not change. Screening strategies are allocated according to breast density at program entry. We predict changes in program sensitivity and tumour size at detection after six years of screening, for a range of targetting strategies. We estimate that under the optimal strategy tested, program sensitivity could improve from 68% to 75% for women with high breast density, with a concomitant reduction from 82% to 77% for women with low breast density, and a program-wide improvement from 74% to 76%. Under that strategy, the proportion of cancers that are small (<15mm) would increase from 64% to 67% for women with high breast density; reduce from 73% to 69% for women with low breast density; and the program-wide figure would not change (68%). We conclude that BreastScreen Australia may be distributed more equitably if women with high breast density are screened more often and women with low breast density are screened less often. We are currently extending ABCSS to model the entire screened population and thereby estimate changes to false-positive rates under alternate screening strategies, which will provide a more complete picture of the potential harms of more intensive screening. ABCSS is also being extended to model alternative screening modalities such as ultrasound and magnetic resonance imaging (MRI), by modifying the sub-model for the probability of detection.
Publication Year: 2009
Publication Date: 2009-01-01
Language: en
Type: article
Access and Citation
Cited By Count: 3
AI Researcher Chatbot
Get quick answers to your questions about the article from our AI researcher chatbot