. Author manuscript; available in PMC: 2014 Aug 24.
Published in final edited form as:Ann Intern Med. 2014 Feb 4;160(3):145. doi:
10.7326/M13-1217 Abstract BackgroundEnormous controversy exists over how often and at what age to implement mammography screening. Given that scientific evidence supports less frequent screening, the cost differences among advocated screening policies should be better understood.
ObjectiveTo estimate the aggregate cost of mammography screening in the United States in 2010 and compare the costs of screening strategies as currently practiced or recommended by professional organizations.
DesignA simulation model was developed to estimate the cost of mammography screening in 2010 as actually practiced. Three screening strategies were then modeled: annual (40–84 year-olds), biennial (50–69 year-olds), and USPSTF guidelines (biennial 50–74 year-olds, personalized <50 and ≥75 year-olds).
SettingUnited States.
PatientsWomen 40–85 years old.
InterventionMammography annually, biennially, or following USPSTF guidelines.
MeasurementsCost of screening per year, using lower bound of costs (Medicare reimbursement).
ResultsThe estimated cost of mammography screening in the US in 2010 under actual practice was $7.8 billion, with approximately 70% of women participating. The simulated cost of screening the target of 85% of women was $10.1 billion for annual screening, $2.6 billion for biennial screening, and $3.5 billion for USPSTF guidelines. The largest drivers of cost (in order) were: screening frequency, percent women screened, cost of a mammogram, percent digital mammography, and percent women recalled.
LimitationsCost estimates and assumptions used in the model were intended to be conservative.
ConclusionsThe cost of mammography varies by almost $8 billion per year based on screening strategy, even using the lower bound of costs. The USPSTF recommended biennial screening based on similar efficacy to annual frequency and would allow substantial resources to be saved. Personalized risk assessment could be used to identify higher-risk women who might benefit from more frequent, younger initiation, and longer duration of screening and prevention. Given that a population-based strategy following USPSTF recommendations costs much less than actual and annual screening practices, resources may be better allocated by screening women biennially and using personalized risk assessment for women younger than 50 years or older than 75 years.
INTRODUCTIONThe frequency and appropriate age to start mammography screening for the detection of breast cancer have been debated in the United States for decades. Controversy intensified after the United States Preventive Services Task Force (USPSTF) recommended a change to biennial mammography on the basis that both annual and biennial screening reduce mortality but with biennial, the negative impacts are reduced.(1) However in the US, there has been resistance to reducing frequency or modifying the age range for mammography. The USPSTF guidelines conflict with professional organizations such as the American Cancer Society, which recommend annual screening from age 40 years and continued regardless of a woman's age as long she is without serious, chronic health problems. Given the broad population mammography serves, it is important to consider the economic impact of the conflicting guidelines.
The USPSTF, in 2009, recommended biennial screening for women between the ages of 50–74, with consideration of screening 40–49-year-old women on a risk benefit decision.(2) The USPSTF recommendations are based on a rigorous review of screening trials and work from the Cancer Intervention and Surveillance Modeling (CISNET) investigators that demonstrated there is little or nothing to gain by increasing the frequency of mammography.(3) The CISNET modeling is corroborated by evidence from the Breast Cancer Surveillance Consortium (BCSC) showing that false-positive recall and biopsy rates are significantly lower in the setting of biennial screening, but without a significant increase in the number of later stage cancers detected.(4–6) The USPSTF recommendations on frequency are now in alignment with most European countries, where many of the defining mammography trials were conducted, with the exceptions of the United Kingdom and Finland, which screen every three years.(7–12)
Screening in the US is delivered locally or regionally and covered by myriad payer and health plan organizations. Thus the total resources required or the cost-tradeoffs of different recommendations are currently unknown. This study was designed to inform the debate by estimating the lower bound of the aggregate cost of mammography screening practiced in the US as compared to recommended screening strategies. Our findings should be valuable to women, clinicians, and health policy makers alike who are aware of the many conflicting guidelines.
METHODS Study DesignTo estimate the cost of mammography in the US, we created a simulation model using mammography screening in 2010 as our base case. We then simulated three strategies (Annual, Biennial, and USPSTF), from the payer perspective. Analyses were performed using R statistical software.(13)
Table 1 shows the four screening strategies, one of which is an estimate of actual practice. (14,15) The other three standardize on the population screened (85%) but differ on the age at which to start and stop and the frequency at which to screen. The biennial strategy represents the European approach, the annual strategy reflects the American Cancer Society (among others) recommendations, and the USPSTF strategy represents a risk-based strategy for screening under 50 and over 75 based on their 2009 recommendations.
Table 1.Model Inputs and Formulas including Mammography Screening Strategies Simulated
Screening Strategy Frequency Modeled Participation 40–49 years 50– 69 years 70– 85 years Annual Strategy Every year 85.0% 85.0% 85.0% Biennial Strategy Every 2 years 0.0% 85.0% 0.0% USPSTF Every 2 years 20.0% (high-risk) 85.0% 25.0%-37.2% Actual Practice* Variable 61%-71% 74%-75% 72%-74% Other Inputs 40– 49 years 50–69 years 70– 85 years Women at Risk (52)** 21994479 36820954 12610766 Percent Recalled†The final output of the model was the aggregate cost of mammography screening per year. The summation included the cost of screening mammograms, the cost of computer-aided detection (CAD), and the cost of recalls and biopsies. A description of the modeling methods is available in Supplemental Appendix A.
Inputs and VariablesModel inputs were attained from multiple sources including the BCSC(16) and are listed in Table 2. (17–22) All input variables except costs were age-specific. The number of mammograms was calculated by determining the population of women at risk by using census data. To focus on screening as opposed to diagnostic or surveillance mammography, we limited the population of women at risk to those between the ages of 40 and 85 and excluded the number of women diagnosed with breast cancer in the past five years, who should be receiving surveillance mammography.
Table 2.Total Cost of Mammography Screening by Strategy in Billions of Dollars
40– 49 years 50– 69 years 70– 74 years >75 years Total CostWe used data from the Behavioral Risk Factor Surveillance System (BRFSS) 2010 Survey, a telephone health survey conducted by the Center for Disease Control, to determine the frequency and percentage of women receiving mammography. We corrected for survey bias by using the correction suggested by Rothman et al.; and though the survey does not distinguish between screening and diagnostic, we excluded women below 40 and above 85 years and those with recent history of breast cancer to best estimate screening as opposed to diagnostic mammograms.(23) In our base case model of actual practice, we included women who reported receiving a mammogram in the past one to five years and estimated the number who would have screened in one given year; otherwise, the simulated strategies only simulated women receiving mammography every one or two years. The simulated strategies modeled an optimal participation of 85%, a screening participation achieved in the past for cervical cancer screening.(24) The USPSTF strategy modeled 20% of 40–50 year-old-women as high-riskting a full and simulated biennial screening for this cohort based on evidence that 40–49 year-old women with a 2-fold increased risk have similar harm-benefit ratios from biennial screening as 50–74 year-old women with average risk.(25) The USPSTF strategy also modeled screening women between the ages of 70 and 85 who are healthy, defined as having fewer than three self-reported chronic conditions as reported by Medicare. (26)
The percent of recalled mammograms was obtained from the BCSC using mammography screening performance data from 1,908,447 exams for 749,597 women screened from 2001–2007. Following the BI-RADS manual,(27) a recall was defined as an initial BI-RADS assessment of 0 (needs additional imaging evaluation), 4 (suspicious abnormality), 5 (highly suggestive of malignancy), or a BI-RADS assessment of 3 (probably benign finding) if it was accompanied by a recommendation for immediate work-up. Separate estimates were computed for recall rates at first and subsequent mammography (i.e. prevalent and incident screens) as well as stratified by frequency of screening, digital versus film mammography, and a woman’s age.
The estimated costs of the modeled strategies include the cost of the screening mammogram and the subsequent recall costs. Costs for mammograms and CAD were determined using 2010 national Medicare reimbursements rates. Recall costs were calculated from the Digital Mammography Imaging Screening Trial’s (DMIST) results of work-up costs including additional imaging and biopsies from false positive and true positive exams.(22) We adjusted DMIST recall costs proportional to the use of digital versus film mammography in 2010. We adjusted all cost data to 2010 US dollars based on inflation as estimated by the medical portion of the Consumer Price Index.(28)
Sensitivity AnalysisWe used Monte Carlo simulations to estimate the uncertainty of our total cost estimates and to quantify the sensitivity of the output (total cost) to the model inputs (for details, see Supplemental Appendix A). (29) In the sensitivity analysis, all terms in the formulas were assumed to be independent and to follow beta distributions as detailed in Appendix A.
RESULTSOur model simulated screening practices in 2010 and estimated the aggregate cost of mammography per year. Three mammography screening strategies advocated by various professional societies under optimal participation rates were also simulated and yielded costs that ranged from $10.1 to $2.6 billion for the most to the least intensive screening strategies (Figure 1).
Figure 1.Comparison of the Costs of Screening Strategies per Year. Each bar represents the total cost of mammography screening per year, demarcating the costs from screening mammograms and the subsequent recalls and biopsies.
The Aggregate Cost of Mammography Screening in the U.SWe estimated the aggregate cost of mammography screening in the US by simulating screening for 40–85-year-old women, which follow a mixture of screening strategies in actual practice. Given the disparate recommendations and the reality of screening in the US, an individual woman who has had a mammogram in a given year may follow annual screening guidelines, biennial guidelines, or be screened irregularly. We modeled participation rates from 61% of women between the ages of 40–45 to 75% of women between the ages of 65–70.(18) Simulating this mix of actual mammography screening practices for women 40 to 85 years-old, we estimated the aggregate cost of mammography screening in 2010 to be in the range of $7.8 billion. Given that Medicare reimbursements rates are at the lower bound of costs, we additionally estimated the impact on aggregate cost per 10% increase in mammography, CAD, and recall. For each 10% increment above medicare costs, the aggregate costs increase by $700 million.
Estimated Cost of Proposed Screening Strategies with Optimal ParticipationThe results of the estimated costs of the annual, biennial, and USPSTF mammography screening strategies are summarized below and in Table 2.
According to our model, annual screening of women between the ages of 40 and 84 is estimated to cost $10.1 billion per year. This strategy follows the guidelines of many policy making medical groups and would increase mammography screening costs by $2.3 billion per year from actual practice. A 10% increase in reimbursements rates increase the aggregate cost by $1 billion.
Biennial screening of 85% of women between the ages of 50–70 is estimated to cost $2.6 billion per year, the least expensive of the strategies. The biennial strategy is $7.7 billion less than annual screening of the 85% of the population aged 40–84 years and $5.4 billion less than what is estimated to be spent in actual practice. Increasing the reimbursements rates by 10% leads to an increase in aggregate cost by $270 million.
The USPSTF strategy simulates biennially screening 50–74 year-old women, and includes screening of high-risk women 40–49 years old and women 75 to 85 years with fewer than three comorbidities. Screening women according to the USPSTF guidelines is estimated to cost $3.5 billion per year. With the same optimal participation rates of 85%, a USPSTF strategy is estimated to cost $6.7 billion less than an annual mammography strategy. The USPSTF strategy covering 85% of the population costs $4.4 billion less than actual practice. For the USPSTF strategy a 10% increase in reimbursements leads to an increase in aggregate cost by $350 million.
Determinants of CostWe used a sensitivity analysis to determine the inputs to which the model was most sensitive; or in other words, which inputs had the most effect on the cost outcomes of the model (Figure 2.) Frequency of screening was the largest driver of cost and the input to which the model was most sensitive. Varying the frequency of screening from biennial to annual had a cost difference of over $7 billion.
Figure 2.Sensitivity Analyses of Mammography Screening Practice and Advocated Screening Strategies. Tornado diagrams depicting sensitivity analyses done for the largest determinants of cost for each strategy. The x-axis depicts the cost of mammography screening per year. The bold line within each tornado diagram is the point estimate of cost for each strategy and the horizontal bars represent the impact on cost the input ranges have in the sensitivity analyses. Frequency is the largest driver of cost as demonstrated by the wide range of cost between least frequent strategy, biennial screening, versus the most frequent strategy, annual screening. The next largest drivers of cost, by decreasing level of impact, were: the variation of the percent women screened, the cost of individual mammograms, percent film versus digital mammography, the number of recalls and the recall cost.
The next largest drivers of costs for the model in decreasing order of impact were: the variation of the percent women screened, percent film versus digital mammography, the cost of individual mammograms, the number of recalls and the recall cost (Figure 2). Each input was varied across a range of likely possibilities. Since digital mammography and CAD are becoming the standard, we estimated the cost of a complete shift to using these methods; though only about $70 more per mammogram this would cost an additional $1.2 billion compared to actual practice (where digital and CAD are used for about 82% and 27% of all mammograms, respectively). The impact of increased costs per mammogram is less for strategies with lower screening frequency.
A sensitivity analysis was done for each screening strategy (Figure 2). The major drivers of cost remained consistent for all four strategies; however, the cost of screening for each strategy, varying the range of inputs, differs greatly. The magnitude of change is billions of dollars for each strategy, but each varies around its estimated overall screening cost per year and there is little overlap in projected costs.
DISCUSSIONOur study shows that mammography screening in the US, as currently implemented, is estimated to cost in the range of $7.8 billion per year. If a USPSTF strategy was implemented to screen 85% of women, we estimate that it would cost $3.5 billion per year with a simulated 15% greater participation rate and cost $4.4 billion less than current practice. If an annual strategy were implemented to screen 85% of the population it would cost an estimated $10.1 billion per year, or $2.3 billion more than current practice. Finally, screening biennially and eliminating screening for women in their forties or over seventy would cost $2.6 billion, not much less than the broader USPSTF strategy.
In 2001, the aggregate cost of mammography was estimated to cost $3–5 billion for all US screening aged women.(30) A study focused on the cost of breast cancer screening for a subset of women, the in fee-for-service Medicare population, used the SEER-Medicare database to estimate screening cost per beneficiary billed during 2006–2007 and multiplied the total number of Medicare beneficiaries to estimate a cost greater than $1billion annually for this subset of older Medicare women.(31) This result is consistent with our analysis considering the use of 10–20% digital mammography during that earlier time period. They also noted that increases in the unit cost of mammography, from digital mammography and CAD, explained 65% of the difference in the regional variation of costs. The introduction of digital mammography confers many advantages including ease of finding prior films, eliminating film processing, and has the opportunity to improve recall rates by eliminating BI-RADS 0 and instituting better biopsy thresholds. The impact of mammography cost changes, such as the increased use of digital, is much less when mammography frequency is lower and is a good investment. Our analysis identifies alternative strategies that will maintain screening performance but can lower the associated burdens by reducing additional recalls and biopsies using fewer resources.
Mammography frequency and age appropriateness affect the cumulative number of abnormal exams that require supplemental imaging, clinical evaluation and possible biopsy. Over the past decade, we have come to realize more of mammography’s limitations.(32,33) Multiple studies have shown that annual mammography does not have improved outcomes for cancer detection as compared to biennial screening.(3,34,35) However, annual compared to biennial screening is associated with a greater likelihood of false-positive recalls (61.3% vs. 41.6%) and biopsies (7% vs. 4.8%) over ten years.(4) Such consequences impact women’s well-being and quality of life.(36,37) The costs per quality adjusted life year (QALY) for annual compared with biennial mammography are more than $340,000 for all ages, well beyond the $75,000-$100,000 per QALY considered a cost-effective intervention.(14) The CISNET modeling group projected that women with a relative risk of 1.9 (range 1.5 to 4.4) for breast cancer were the women likely to benefit from biennial screening in their forties.(25) To determine the appropriate age to start screening, experts recommend risk-based approaches for personalized screening, which is more in line with USPSTF guidelines.(38)
The implementation of the Affordable Care Act is expected to increase screening by 500,000 mammograms per year as mammography access improves.(39,40) Disparities in mammography relate to access to high-quality facilities and use of breast imaging specialists to read mammograms.(41,42) Mammography quality is improved when read by dedicated mammographers at high quality facilities, which would also likely lower recalls, a large driver of screening costs.(43) Recall rates vary significantly nationally and internationally and the average recall rates in the US are double to triple of those reported in Europe.(44–47) Biennial screening policies that target age-appropriate women make high-quality mammography available to more women while reducing their travel and time burdens.
The Institute of Medicine report Best Care at Lower Cost describes opportunities to reduce costs and improve the value of health care including missed prevention opportunities ($55 billion) and inefficiently delivered services ($130 billion).(48) Tailored screening policies address improved care at lower cost. As our analysis suggests, similar or higher quality screening can be obtained at similar or lower costs, freeing resources for risk-assessment and prevention that are not currently part of screening. Learning to set better thresholds for recall and biopsy would also safely generate substantial savings and reduced burden on women. (49,50)
Limitations of this analysis include the mammography participation estimates in which we were conservative by using a correction to the BRFSS and used rates lower than the CDC’s estimates of 72–79%.(51) We used Medicare costs, which are at the lower end of reimbursement rates and thus actual costs could be substantially higher. Recall costs were inflated to 2010 costs from 2005 data; however, current diagnostic workups may now be more expensive with greater use of imaging technology such as MRI. We did not evaluate direct nonmedical or indirect costs associated with screening such as transportation or loss of time off from work. The additional cost of digital mammography compared to film was only modeled for the actual costs of the mammogram and not associated costs. We did not model the overall costs of a missed diagnosis; though evidence suggests that biennial screening is not associated with a statistically significant increase in the diagnosis of late stage cancers nor that screening costs make a significant difference in the costs of initial cancer treatment.(4,6,31) We did not address the impact of possible overdiagnosis on costs.(32,50) Mammography practices may have changed since 2010 to align actual practice with USPSTF guidelines; however, recent studies have shown that since the release of the guidelines, screening practices have not greatly changed.(52) Lastly, we excluded the cost of screening women under 40 and older than 85 to simplify guidelines and focus on screening; however, screening women outside of recommended policies is likely to have similar costs across strategies. The accumulation of assumptions is likely to make the projected costs an underestimate for each strategy. This analysis was not intended to provide a highly accurate point estimate; but rather, an approximate range and relative ranking of the costs for each strategy.
Although an individual mammogram is not expensive, the aggregate national cost of screening millions of women is enormous. Costs of false-positive recalls and biopsies are also magnified and contribute to making mammography so resource intensive. Better understanding of the limitations of mammography will enable using this screening tool more effectively. A screening policy following the USPSTF guidelines uses fewer resources, has fewer false-positive biopsies and recalls compared to annual screening and is being incorporated into quality guidelines.(53) Those who advocate annual screening should justify the increased costs of almost $7 billion per year compared to biennial policies.
The billions saved from avoiding less-effective mammography screening can be used in multiple ways to improve women’s health including: increasing participation in screening; routine assessment of breast cancer risk and referral for breast cancer prevention services if at high risk, especially for women less than 50 years; genetic counseling for women with strong family history (52); alternative modalities for screening women with dense breast tissue; and supporting a more programmatic/ public health approach to screening (as in Europe) with double reading, screening by invitation, and outcomes tracking. Resources can be better allocated to improve screening, with the emphasis on higher quality mammograms read by specialized mammographers (54). More resources should go into improving the quality of the reads rather than simply a higher frequency. It is also clear that higher unit reimbursement costs have a lower impact on aggregate costs when the frequency is less, such as with the USPSTF recommendations. Numerous publications have demonstrated that the most experienced readers, or systems of double reading, improve sensitivity as well as specificity.(55) As our sensitivity analysis demonstrates, the cost per mammogram is not the largest driver of cost, but rather frequency. High quality interpretation should be covered, as it would improve screening performance and reduce recall costs. Lastly, current screening participation rates fall short of national goals(56) and though the Affordable Care Act aims to improve access, utilizing mammography effectively following guidelines such as the USPSTF will ensure that resources exist to expand screening participation.
In conclusion, mammography screening is resource intensive at an estimated cost of $7.8 billion per year as currently practiced. Following mammography screening guidelines, such as the USPSTF, that optimize frequency on the basis of best available evidence will put us in a position to improve screening and save billions of dollars that can be invested in personalized risk-based screening and prevention strategies. Such a change in screening practice is likely to improve the quality of screening and is in line with our national goals of advancing health care delivery while improving cost-efficiency.
Supplementary MaterialArrays
ACKNOWLEDGEMENTSThe collection of cancer data used in this study was supported in part by several state public health departments and cancer registries throughout the US For a full description of these sources, please see: http://www.breastscreening.cancer.gov/work/acknowledgement.html.
The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Cancer Institute or the National Institutes of Health. We thank the BCSC investigators, participating women, mammography facilities, and radiologists for the data they have provided for this study. A list of the BCSC investigators and procedures for requesting BCSC data for research purposes are provided at: http://breastscreening.cancer.gov/.
Thank you to Pamela Derish, MA, in the Department of Surgery at UCSF for editing this manuscript and to Karla Kerlikowske, MD, Department of Medicine and Epidemiology/Biostatistics at UCSF, and Rebecca Hubbard, PhD, Group Health Research Institute and Department of Biostatistics at University of Washington for reviewing the manuscript and for giving valuable input.
GRANT SUPPORT
This effort is part of the analysis proposed for breast cancer screening and prevention as part of the Athena Breast Health Network, a collaboration of the five University of California medical centers. Athena is supported by the University of California and Safeway Inc.
Data collection for this work was supported by a National Cancer Institute-funded Program Project (P01CA154292) and the Breast Cancer Surveillance Consortium (U01CA63740, U01CA86076, U01CA86082, U01CA63736, U01CA70013, U01CA69976, U01CA63731, U01CA70040, HHSN261201100031C). Each registry and the Statistical Coordinating Center (SCC) have received institutional review board approval for either active or passive consenting processes or a waiver of consent to enroll participants, link data, and perform analytic studies. All procedures are Health Insurance Portability and Accountability Act (HIPAA) compliant and all registries and the SCC have received a Federal Certificate of Confidentiality and other protection for the identities of women, physicians, and facilities who are subjects of this research.
Dr. O’Donoghue funded in part by the University of Illinois at Chicago Eleanor B. Pillsbury Fellowship.
Dr. Eklund is funded by the Swedish Council for Working Life and Social Research (FAS), project number 2012–0073.
Primary Funding Source: Athena Breast Health Network supported by the University of California and Safeway Inc.
FootnotesNone of the authors have any conflicts of interest.
All authors had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Contributor InformationCristina O’Donoghue, Department of Surgery, University of California at San Francisco, San Francisco CA; Department of Surgery, University of Illinois at Chicago, Chicago, IL.
Martin Eklund, Department of Surgery, University of California at San Francisco, San Francisco CA.
Elissa M. Ozanne, Department of Surgery and the Institute for Health Policy Studies, University of California at San Francisco, San Francisco, CA.
Laura J. Esserman, Department of Surgery and the Institute for Health Policy Studies, University of California at San Francisco, San Francisco, CA.
ReferencesThis section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary MaterialsArrays
RetroSearch is an open source project built by @garambo | Open a GitHub Issue
Search and Browse the WWW like it's 1997 | Search results from DuckDuckGo
HTML:
3.2
| Encoding:
UTF-8
| Version:
0.7.3