Archives

  • 2019-10
  • 2019-11
  • 2020-03
  • 2020-07
  • 2020-08
  • Another important contribution of our study is that

    2020-08-12

    Another important contribution of our study is that our model explored screening strategies that are based on specific guideline recommendations instead of hypothetical scenarios. This improves 
    the practical implications LDN-193189 of our study finding because clinicians and payers often follow guideline-recommended screening practices. As mentioned previously, two unique features of the updated ACS guideline are the starting age of screening and the hybrid interval. Although the CISNET group has explored hybrid strategies, these strategies differ from those in the updated ACS guideline in that screening transitioned to biennial intervals at the age of 50 years rather than 55 years. The transition from annual screening intervals to biennial screening intervals at the age of 55 years is an important screening strategy to consider because a recent analysis of the Breast Cancer Surveillance Consortium data showed that among pre-menopausal women, those with biennial screening intervals were more likely to have less favorable tumor characteristics (e.g., larger tumor size or tumor associated with poor prognosis at diagnosis) than those with annual screening intervals [59]. The age for tran-sitioning from annual to biennial intervals in the updated ACS guideline was based on the rationale that most women would be postmenopausal by the age of 55 years, thus preserving the benefits of annual screening for premenopausal women [5]. Indeed, when we included transitioning at the age of 55 years in our model, the cost-effectiveness comparison favored the updated ACS guideline.
    As screening technology advances, it is important that clinical parameters in modeling studies reflect the standard of care in current practice. Although the LDN-193189 of digital breast tomo-synthesis (also known as three-dimensional mammography) with conventional two-dimensional digital mammography has been shown to have better breast cancer detection rates and lower false-positive recalls than digital mammography alone [60], digital mammography is still considered the current standard of care for breast cancer screening among average-risk women. Therefore, our model was based on the performance characteristics of digital mammography. From a modeling perspective, it is feasible to switch from one screening modality to another over time in anticipation of digital breast tomosynthesis or another new mo-dality eventually becoming the standard of care. Nevertheless, without solid evidence as to when the transition to a new standard of care will occur or when the performance characteristics of a potential screening modality will become available in the future,
    Fig. 3 – Cost-Effectiveness Acceptability Frontier of Breast Cancer Screening Strategies. Abbreviations: AAFP, American Association of Family Physicians; ACOG, American Congress of Obstetricians and Gynecologists; ACP, American College of Physicians; ACR, American College of Radiology; ACS, American Cancer Society; QALY, quality-adjusted life-year; USPSTF, US Preventive Services Task Force.
    such an analysis would be highly speculative and could create major challenges for the interpretation of findings from the CEA of screening strategies that involve different initiation ages and screening intervals because of the intermixing of screening stra-tegies and screening modalities. We use the transition from film to digital mammography to illustrate this point. Most screening fa-cilities had replaced film mammography with digital mammog-raphy by 2010. Therefore, for the 1960 birth cohort on which our model was based, by the time the women in this cohort reached the age of 50 years, most of them would have started breast cancer screening with digital mammography. Nevertheless, for screening strategies with the initiation age of 40 years, some women in our simulation model would have started with film mammography but eventually would have been switched to digital mammography as they aged into their 50s. If we were to include changes in screening modalities over time in our model, it would not be possible to determine whether the estimated difference between screening strategies with the initiation age of 50 years versus 40 years was driven by the difference in the initiation age or in the test characteristics between screening modalities. Therefore, it is our view that for modeling studies that compare screening strategies involving different initiation ages and screening intervals, it is preferable to base the model on one screening modality so that findings are more transparent and easier to communicate to policymakers.
    This study has several limitations. First, we assumed 100% compliance for each screening schedule, which does not reflect the actual screening pattern at the population level. Nevertheless, because our focus is to compare the relative benefits and costs between different screening strategies, making such comparisons under the scenario of optimal compliance is a reasonable modeling approach. Second, we did not consider the disutility associated with false-positive screening test results, which should have a relatively small impact on QALYs, as shown in CISNET models [57]. This finding is further confirmed in a recent survey study that evaluated the impact of false-positive mammograms on women’s anxiety, health utility, and attitudes toward future screening [61]. The authors of the study concluded that although false-positive mammograms were associated with increased short-term anxi-ety, they were not associated with a measurable health utility decrement. Finally, although the use of ICER allows for a compar-ison across different strategies, relying on point estimates of ICER to determine the cost-effectiveness of screening strategies is subject to uncertainty because some screening strategies yield very small differences in effectiveness. We thus also used probabilistic anal-ysis to show which strategy has the highest probability of having the largest expected net benefit against different levels of societal willingness to pay. Compared with one-way sensitivity analyses seen in many modeling studies, this approach has the advantage of capturing the combined effect of the uncertainty of all parameters.