The estimated impact of COVID-19 on population breast cancer screening outcomes, and options for risk-based recovery

ABSTRACT

OBJECTIVES AND RATIONALE Estimating the impact of COVID-19 on cancer screening programs and related outcomes can help health services prepare for potential delays in diagnoses and different demands on treatment services and plan for best approaches to recovery. Simulation modelling enables estimation of outcomes for a range of scenarios. In this study, we estimate the impact of various disruptions and recovery strategies for the Australian biennial mammographic breast screening program (BreastScreen). METHOD Policy1-Breast is a continuous-time, multiple-cohort micro-simulation model that simulates the whole Australian female population, incorporating breast cancer risk and natural history, breast density, menopause, hormone therapy use and breast cancer screening. Firstly, in the early stages of the COVID pandemic we used Policy1-Breast to evaluate how 3, 6, 9 and 12-month pauses to BreastScreen would impact on population-level breast cancer diagnoses, tumour staging, and breast cancer survival, compared to business-as-usual (BAU) outcomes. Secondly, to explore options for recovery after an actual one-month screening pause in April 2020, we evaluated a range of assumed throughput levels following screening resumption (50% or 80% up to December, then 100% to 120% from Jan 2021), comparing various protocols where specific sub-groups of clients were prioritised for screening during the recovery Speriod. Outcomes are reported for the target age range for the BreastScreen program (50-74 years). RESULTS For 3-to 12-month pauses, we estimated a slight reduction in 5-year survival following diagnosis for women directly affected by a pause, but no discernible changes to population-level breast cancer mortality rates up to 2023. We estimated marked fluctuations in population rates of invasive breast cancer diagnoses with a 10% increase in cancer diagnoses between 2020-2021 and 2022-2023. For a 12-month pause to screening we estimate that population-level breast cancers would increase in size (with an additional 4% >15mm at diagnosis) and be more likely to involve the nodes (increasing from 26% to 30% of all cancers). We estimate that median screening intervals during 2020-2021 would increase from 104 weeks under BAU up to 130 weeks with a 12-month pause, and BreastScreen recall rates and false positive recall rates would fluctuate markedly over time. For the second evaluation of a one-month pause followed by various throughput and prioritisation scenarios, we estimated that screen-detected cancer rates would vary markedly with throughput but interval cancer rates would not, leading to fluctuations in program sensitivity of up to 6%. Reflecting the periodic nature of screening participation, we estimated the extent to which longer-term future screening participation rates are expected echo the peaks and troughs in participation due to the impacts of the COVID pandemic in 2020. We estimate that for a given throughput assumption, client prioritisation could lead to different rescreening rates, screening intervals, and time required for prioritisation protocols, with little change to cancer outcomes. CONCLUSION These modelled evaluations estimate short and longer-term effects of COVID-19 on the impact of population breast cancer screening in Australia. The estimated changes in breast cancer rates and characteristics would be expected to have a flow-on effect on the demand for treatment services in terms of throughput and case-mix. Preparing for such outcomes is critical given that treatment services are also directly impacted by the pandemic. The modelled outcomes are likely to be relevant to other high-income settings with established population breast cancer screening programs.