br In the United States
In the United States, for women at an average risk of devel-oping breast cancer, several specialty societies, including the US Preventive Services Task Force (USPSTF), the American Cancer Society (ACS), the American College of Radiology (ACR), the American Congress of Obstetricians and Gynecologists (ACOG), the American Association of Family Physicians (AAFP), and the American College of Physicians (ACP), have published breast cancer screening guidelines. Both the ACS and the USPSTF updated their breast cancer screening guidelines in late 2015/early 2016. The 2016 updated guideline from the USPSTF was similar to its recommendation in 2009 , but the 2015 updated ACS guide-line differed substantially from its previous guideline published in 2003 . Two important differences between the updated ACS breast cancer screening guideline and its 2003 guideline (which recommended annual mammography screening starting at the age of 40 years) are that 1) the recommended starting age of screening is 45 years and 2) the recommended screening interval transition from annual to biennial is after reaching the age of 55
* Address correspondence to: Ya-Chen Tina Shih, Section of Cancer Economics and Policy, Department of Health Services Research, The University of Texas MD Anderson Cancer Center, 1515 Holcombe Boulevard, Unit 1444, Houston, TX 77030.
E-mail: [email protected]
1098-3015/$36.00 - see front Rituximab Copyright © 2019, ISPOReThe Professional Society for Health Economics and Outcomes Research. Published by Elsevier Inc.
years. Similar to the updated USPSTF guideline, the updated ACS guideline does not suggest a specific age to stop screening; instead it offers a general guidance that mammography screening should be continued as long as a woman’s overall health is good with a life expectancy of 10 years or more, leaving the decision to the discretion of women and their physicians as in the case for women aged between 40 and 44 years. The scientific rationale behind these recommendations is influenced by the quality of existing evidence and the consideration of trade-off between benefits and harms related to screening mammography .
The objective of our study was to use a microsimulation model to assess the cost-effectiveness of three US-based mammography screening guidelines: the updated and previous ACS guidelines and the updated USPSTF guideline. It should be noted that this comparison essentially covers guidelines from other specialty societies because the AAFP and ACP guidelines agree with the updated USPSTF guidelines, and the ACOG and ACR guidelines are similar to the previous ACS guideline that recommends annual mammography screening starting at the age of 40 years. Findings from this assessment will assist deci-sion makers in planning, designing, and promoting efficient cancer control strategies for women at an average risk of devel-oping breast cancer.
Natural History Model
We generated the natural histories of a large cohort of women at an average risk of developing breast cancer by incorporating ductal carcinoma in situ (DCIS) into a previously published microsimulation model for invasive breast cancer [6,7]. The sta-tistical details for the construction of the microsimulation model can be found in the study by Shen and Parmigiani , and the model structure is described in the schematic diagram in the Supplemental Materials found at https://doi.org/10.1016/j.jval.2 018.07.880. Briefly, the model simulated each woman’s natural history throughout her lifetime for a large cohort of women born in 1960 using Monte-Carlo simulations. For women who devel-oped breast cancer according to their age-specific incidence , the model captured four states of disease progression: disease-free or asymptomatic, screen-detectable preclinical disease, clinical disease, and death due to competing risks or breast can-cer. Because the ages at onset of preclinical disease are unob-servable, sporangia were derived from age-specific breast cancer incidence and the preclinical sojourn time distributions. A random preclinical sojourn time was generated for each woman depending on her age at onset of preclinical disease, with uncer-tainty incorporated through an inverse gamma prior for the mean sojourn time using parameters that match the estimated means and SDs from the screening trials .
Of the two forms of DCIS (progressive and nonprogressive), our model focused explicitly on nonprogressive DCIS because treat-ments received by patients with this condition represent over-diagnosis and overtreatment. Specifically, we assumed that nonprogressive DCIS remains indolent and/or regresses to a normal state for women with this diagnosis; therefore, they would not be clinically diagnosed if the DCIS was not detected from screening mammography, nor would they die of DCIS. The pro-gressive form of DCIS would be captured as invasive breast cancer under this modeling approach. We calibrated the percentage of nonprogressive DCIS to be about 9% using data provided by van Ravesteyn et al. , who reported that approximately 17% of patients with breast cancer were diagnosed with DCIS, and of those, 51.3% had nonprogressive DCIS.