Table 4.

Summary of health economic and outcomes models evaluating risk-stratified breast screening identified during the scoping review.

Study reference	Population modelled	Modelling approach utilised	Risk groups and screening scenarios simulated	Key results
Sankatsing et al. [116]	Netherlands—women without BRCA1/2 mutation Women born in 1974 Time horizon age 40–death	Microsimulation (MISCAN—microsimulation screening analysis; semi-Markov processes)	Risk groups: low, average and high, using ‘common risk factors’, excluding breast density Simulated: biennial screening aged 50–74 overall; biennial or triennial screening for low-risk women starting 50–60 to 64–74 years; annual or biennial screening for high-risk women starting 40–50 until 74–84 years Assumption of 100% screening attendance	Per 1000 women: Biennial screening to all, 50–74: 206 life-years gained, 16 deaths avoided, 187 false positives, 5 overdiagnosed cases Triennial screening 50–71 for low risk: 134 life-years gained, 10 breast cancer deaths avoided, 102 false positives, 3 overdiagnosed cases Biennial screening 40–74 for high risk: 380 life-years gained, 26 breast cancer deaths avoided, 371 false positives, 7 overdiagnosed cases
Arnold et al. [114]	Germany Women aged 50, followed to age 100 or death	Microsimulation Markov model	Risk factors: family history, personal history of biopsy, breast density Compared annual, biennial and triennial universal screening to risk-adapted strategies based on relative risk (three risk categories) Assumption of 54% adherence	Risk-stratified programmes may be more efficient, depending on mortality reduction or QALYs are strategic focus At 54% adherence, compared with no screening, screening women with relative risk >1 was projected to generate 8.63% mortality reduction, incremental QALYs of 0.023 and incremental costs of 211 Euros per woman (2017 prices).
Sun et al. [117]	China (urban population)	Prior natural history Markov model	High-risk defined: relative risk >2 High-risk women: screened using USS aged 40–44 years with subsequent mammography if indicated; with both modalities if 45–69 years Low-risk women: no screening (diagnosis after symptoms arise) Simulated complete treatment and also 70% treatment after diagnosis	Compared with no screening, risk-adapted approach screening every 3 years, with full treatment: Lifetime costs US$184 per case (2014 prices), 22.99 QALYs, 0.0127 difference in QALY Compared with no screening, annual screening and full treatment: Lifetime costs US$335.43 per case (2014 prices), 23.01 QALYs, 0.028 different in QALYs
Pashayan et al. [16]	UK Cohort n = 364,500 Women aged 50 years followed up to 85 years	Life-table model	Three cohorts with screening based on risk group: (1) No screening (2) Screen all women aged 50–69 years as per NHS BSP (3) Only women above risk polygenic risk threshold screened every 3 years from age 50–69 years	Compared to no screening, risk-based screening: Overdiagnosis:deaths prevented ratio increased from 0.07 to 0.99 as risk threshold lowered (from 99th to 71st percentile) Minimum ICER at 77th percentile of risk threshold (£11,911 per QALY gained), versus £66,445 when using 99th percentile as risk cut-off (price date not specified) At 32nd percentile of risk, risk-adapted screening generated an incremental cost of £20,066 (price date not specified), 450 more QALYs, and 7 fewer breast cancer deaths
Gray et al. [111]	UK Women eligible for NHS BSP (aged 50–70 years)	Discrete event simulation	Four stratification methods in NHS BSP: (1) Absolute 10-year risk as per of Brentnall (BCR 2015): <3.5% triennial screening, 3.5–8% biennial screening, >8% annual screening (2) Relative 10-year risk: low tertile = triennial, middle tertile = biennial, high tertile = annual (3) Supplemental US for women with high breast density (4) Approach 1 plus the supplemental US as in (3)	Compared to the current NHS BSP strategy (screening 50–70 years of age ever 3 years): Risk-stratification methods 1 and 2 were deemed cost-effective relative to threshold range of £20,000–30,000 per QALY ICER for method 1 versus UK BSP: £16,689 (2015 prices) ICER for method 2 versus UK BSP: £23,924 (2015 prices)
Trentham-Dietz et al. [112]	US Women aged 50+ Lifetime horizon	Microsimulation models ×3	Examined various combinations of breast density (four categories) and relative risk for factors other than density (1.0, 1.3, 2.0 and 4.0) Settings of annual/biennial/triennial screening for women aged 50–74 years, and also for women 65–74 years Assumed 100% adherence	Per 1000 women with fatty breasts/scattered fibroglandular density + RR 1 or 1.3: Biennial screening (50–74): 5.1 deaths averted Triennial screening (50–74): 3.4 death averted Biennial screening (50–74): 4.1 deaths averted Triennial screening (65–74): 6.5 deaths averted Triennial screening for average-risk women with low-density breasts provided favourable balance of harms and benefits and is cost-effective Annual screening for higher risk (RR 2.0 or 4.0) with heterogeneously or very dense breasts has favourable balance of benefits and harms and is cost-effective
Schousboe et al. [113]	US Women aged 40–49, 50–59, 60–69 and 70–79 (initial mammography at 40) Lifetime horizon	Markov cost–utility microsimulation model	Modelled risk based on BI-RADS breast density category, and up to 2 risk factors (family history or previous biopsy) Examined annual, biennial, triennial, 3–4 yearly mammography or no mammography	A range of cost-effective strategies for women of different age groups, breast density and presence of up to 2 risk factors were identified (assuming $100,000 and $50,000 cost-effectiveness thresholds), e.g. at a 50,000 cost-effectiveness threshold: BI-RADS B-D, or BI-RADS A + 1–2 risk factors: biennial screening 50–59 years, reassess at age 60 BI-RADS A + 0/1 risk factors: 3–4 yearly mammography 50–59 years, reassess at age 60

BI-RADS Breast Imaging-Reporting and Data system classification, USS ultrasound scan, NHS BSP National Health Service Breast Screening Program (in the United Kingdom), ICER incremental cost-effectiveness ratio, QALY quality-adjusted life year.