Fax: (617) 726-9414
Cost-effectiveness of alternating magnetic resonance imaging and digital mammography screening in BRCA1 and BRCA2 gene mutation carriers
Article first published online: 26 NOV 2012
Copyright © 2012 American Cancer Society
Volume 119, Issue 6, pages 1266–1276, 15 March 2013
How to Cite
Cott Chubiz, J. E., Lee, J. M., Gilmore, M. E., Kong, C. Y., Lowry, K. P., Halpern, E. F., McMahon, P. M., Ryan, P. D. and Gazelle, G. S. (2013), Cost-effectiveness of alternating magnetic resonance imaging and digital mammography screening in BRCA1 and BRCA2 gene mutation carriers. Cancer, 119: 1266–1276. doi: 10.1002/cncr.27864
- Issue published online: 4 MAR 2013
- Article first published online: 26 NOV 2012
- Manuscript Accepted: 25 SEP 2012
- Manuscript Revised: 18 SEP 2012
- Manuscript Received: 5 AUG 2012
- BRCA1 gene;
- BRCA2 gene;
- breast neoplasms;
- cancer screening;
- cost effectiveness
Current clinical guidelines recommend earlier, more intensive breast cancer screening with both magnetic resonance imaging (MRI) and mammography for women with breast cancer susceptibility gene (BRCA) mutations. Unspecified details of screening schedules are a challenge for implementing guidelines.
A Markov Monte Carlo computer model was used to simulate screening in asymptomatic women who were BRCA1 and BRCA2 mutation carriers. Three dual-modality strategies were compared with digital mammography (DM) alone: 1) DM and MRI alternating at 6-month intervals beginning at age 25 years (Alt25), 2) annual MRI beginning at age 25 years with alternating DM added at age 30 years (MRI25/Alt30), and 3) DM and MRI alternating at 6-month intervals beginning at age 30 years (Alt30). Primary outcomes were quality-adjusted life years (QALYs), lifetime costs (in 2010 US dollars), and incremental cost-effectiveness (dollars per QALY gained). Additional outcomes included potential harms of screening, and lifetime costs stratified into component categories (screening and diagnosis, treatment, mortality, and patient time costs).
All 3 dual-modality screening strategies increased QALYs and costs. Alt30 screening had the lowest incremental costs per additional QALY gained (BRCA1, $74,200 per QALY; BRCA2, $215,700 per QALY). False-positive test results increased substantially with dual-modality screening and occurred more frequently in BRCA2 carriers. Downstream savings in both breast cancer treatment and mortality costs were outweighed by increases in up-front screening and diagnosis costs. The results were influenced most by estimates of breast cancer risk and MRI costs.
Alternating MRI and DM screening at 6-month intervals beginning at age 30 years was identified as a clinically effective approach to applying current guidelines, and was more cost-effective in BRCA1 gene mutation carriers compared with BRCA2 gene mutation carriers. Cancer 2013. © 2012 American Cancer Society.
Women who carry breast cancer susceptibility gene mutations (BRCA1 or BRCA2) are at increased risk of developing breast cancer, and approximately 45% to 65% are diagnosed with breast cancer by age 70 years.1 Multiple guidelines for this high-risk population recommend earlier breast cancer screening with both mammography and magnetic resonance imaging (MRI). However, there is variability across recommendations and in their clinical application. The National Comprehensive Cancer Network recommends screening beginning at age 25 years,2 whereas the American College of Radiology recommends starting at age 30 years.3 The American Cancer Society does not specify an age, recommending that the decision about when to initiate screening should be based on individual circumstances and preferences.4
Current guidelines also do not specify how to combine mammography and MRI. Published trials of MRI and mammography screening for women who have an increased risk of breast cancer have used both tests within 90 days of each other, usually on the same day.5 More recent studies indicate that MRI can detect cancers that are not identified on mammography 6 months earlier6 and suggest that alternating MRI and mammography every 6 months may provide greater life-expectancy gains than contemporaneous dual-modality screening.7
Diagnostic imaging use, including MRI, has increased substantially over time,8, 9 and imaging cost increases have outpaced the rate of total cost increases in Medicare beneficiaries.9 Concern over rising health care costs has prompted an increased focus on providing high-value, cost-conscious care.10, 11 This approach involves assessing both benefits and harms of diagnostic testing to determine whether patient outcomes have been improved and also examining the costs of care, including costs and savings of downstream testing, treatment, and follow-up.
We have developed a computer simulation model to project long-term health outcomes and costs of screening women with BRCA gene mutations.7, 12, 13 Our results from a previous comparative effectiveness analysis suggest that screening with MRI and digital mammography (DM) at alternating 6-month intervals provided additional survival benefit beyond contemporaneous dual-modality regimens.7 Of 26 strategies that were evaluated, the 3 screening strategies that had the greatest breast cancer mortality reduction were: 1) DM and MRI at alternating 6-month intervals starting at age 25 years, 2) annual MRI starting at age 25 years with alternating DM added at age 30 years, and 3) alternating DM and MRI starting at age 30 years. In the current study, we extended the model to evaluate the cost-effectiveness of these breast cancer screening strategies, which alternate MRI with DM at 6-month intervals.
MATERIALS AND METHODS
No human subject data from individual patients were used for this study; therefore, institutional review board approval was not required.
The Markov Monte Carlo simulation model was programmed in C++ (Microsoft Corporation, Redmond, Wash) and included breast cancer development, detection, and treatment in asymptomatic BRCA1/BRCA2 mutation carriers beginning at age 25. Model input parameters were obtained through a critical review of the published literature and model calibration.7, 12, 13 The cumulative incidence of breast cancer to age 701 in the absence of screening was calibrated14 separately for BRCA1 and BRCA2 mutation carriers, with best fitting natural history parameters previously reported.7, 13 In the base-case analysis, we assumed that women had not undergone prophylactic salpingo-oophorectomy, mastectomy, or chemoprevention. For each screening scenario, 2 million individual women were tracked until death, and their outcomes were aggregated to provide cohort estimates of projected outcomes.
Screening Strategies Evaluated
We examined 3 dual-modality screening strategies that were consistent with current guidelines2-4 (Fig. 1): 1) DM and MRI alternating at 6-month intervals beginning at age 25 years (Alt25), 2) annual MRI at age 25 years with alternating DM added at 6-month intervals beginning at age 30 years (MRI25/Alt30), and 3) DM and MRI alternating at 6-month intervals beginning at age 30 years (Alt30). The following additional comparator strategies were evaluated for calculation of incremental cost-effectiveness: annual DM beginning at age 25 years (DM25), annual DM beginning at age 30 years (DM30), and clinical surveillance without imaging.
Breast Cancer Diagnosis, Treatment, and Mortality
The BRCA1 and BRCA2 cohorts were evaluated separately. Key input parameters for breast cancer detection and diagnosis are presented in Table 1. We assumed that all asymptomatic women underwent screening with perfect adherence. At screening, women received either positive or negative results based on the sensitivity and specificity of either DM7, 15, 16 or MRI.17 Women who had positive screening results underwent further diagnostic workup. Women whose diagnostic workup results were suspicious for breast cancer subsequently underwent biopsy to establish a final diagnosis of malignant or benign disease. If the biopsy results were benign, then the woman was tracked as having both a false-positive screening test and a false-positive biopsy recommendation. Women who had negative screening results underwent no further intervention until the next screening event. If a cancer was missed on a screening test, then cancer progression continued until the next screening event or until the cancer presented clinically as an interval cancer.
|Parameter||Base-Case Value||Range for Sensitivity Analysis||Reference(s)|
|Digital mammographya||Lowry 2012,7, Pisano 2004,15 Pisano 200816|
|Invasive breast cancer||.50||.65|
|MRIa||Leach 2005,17 Kreig 2004,18 Kuhl 2005,19 Sardanelli 2007,20 Warner 200421|
|Invasive breast cancer||.86||.81-.93|
|Radiation exposure risk|
|Radiation dose from screening mammography (mGy, bilateral examination)||4.15||1.7-6.8||Hendrick 201022|
|Excess relative risk model for radiation exposure||Age at exposure||Attained age||Preston 200223|
|Probability of biopsy recommendation in a woman without cancer after a positive screening test||.123||.11-.15||Sickles 2005,24 Rosenberg 200625|
|Probability that a suspicious lesion requires a surgical biopsy||.253||.10-.37||Gutwein 2011,26 Clarke-Pearson 200927|
|Probability of ER-positive breast cancer|
|BRCA1||.10||.22||Lakhani 2002,28 Mavaddat 201229|
|BRCA2||.66||.77||Lakhani 2002,28 Mavaddat 201229|
|BRCA1||.77||.51||Mavaddat 2012,29 Allred 200230|
|BRCA2||.77||.91||Mavaddat 2012,29 Allred 200230|
|Probability of HER2-receptor positive breast cancer||.03||b||Honrado 200631|
In women who were diagnosed with breast cancer, primary tumor diameter at diagnosis and the method of cancer detection (screen-detection or clinical presentation) determined the probability of lymph node involvement and distant metastases.32 Once breast cancer was staged and treated, annual mortality from breast cancer was based on a woman's age at diagnosis, disease stage, and tumor estrogen-receptor status.33 Competing mortality risks were obtained from United States life tables34 and were adjusted to reflect the increased mortality rate from ovarian cancer in BRCA mutation carriers.35 We accounted for mammography-induced breast cancer risks using an excess relative-risk model that depended on a woman's age at exposure, in which the risk of radiation-induced cancer was proportional to the population's underlying breast cancer risk.7, 23
Costs of screening and diagnosis were obtained from the 2010 Medicare Physician Fee Schedule.36 Direct medical treatment costs and patient time costs were obtained from the published literature and were adjusted to 2010 US dollars using the medical care component of the Consumer Price Index37 (Table 2).
|Screening and Diagnostic Costs||Description Code(s)a||Base Case Value, 2010 US $||Range for Sensitivity Analysis, $||Reference(s)|
|Bilateral digital screening mammography||G0202, 77051||144||72-288||2010 Medicare Physican Fee Schedule36|
|Bilateral screening MRI||77059, C8908, 77052||619||310-1238|
|Unilateral digital diagnostic mammogram||G0206, 77052||136||b|
|Percutaneous biopsy, ultrasound-guided||19102, 19295, 88305, 76942, G0206||864||b|
|Percutaneous biopsy, stereotactic||19103, 77031, 19295, 76098, 88305, G0206||1543||b|
|Percutaneous biopsy, MRI-guided||77021, 19103, 19295, 88305, G0206||1517||b|
|Excisional biopsy||19125, 19290, 77032, 76098, 88307||2715||b|
|Sentinel lymph node biopsy||38792, 38525, 88307, 1 mCi Tc99m sulfur colloid||2605||b||(and institutional costs)|
|Annual Stage-Specific Treatment Costsc||Initial Phase: Year 1, $||Continuing Phases: Years 2-5/ Year ≥6, $||Range for Sensitivity Analysis, $||Reference(s)|
|Stage, ER status, age at diagnosis||b||Derived from Oestreicher 2005,38 Mandelblatt 2005,39 Red Book 200740|
|DCIS, ER−, age ≥50 y||9243||1292/1292|
|DCIS, ER−, age <50 y||8931||1292/1292|
|DCIS, ER+, age ≥50 y||10,286||5468/1292|
|DCIS, ER+, age <50 y||9181||2292/1292|
|Local, ER−, age ≥50 y||13,795||1028/1028|
|Local, ER−, age <50 y||12,697||1028/1028|
|Local, ER+, age ≥50 y||11,429||5200/1028|
|Local, ER+, age <50 y||10,315||2024/1028|
|Regional, ER−, age ≥50 y||21,345||1828/1828|
|Regional, ER−, age <50 y||25,940||1828/1828|
|Regional, ER+, age ≥50 y||22,388||6000/1828|
|Regional, ER+, age <50 y||26,190||2828/1828|
|Distant stage (includes chemotherapy, wage loss, and informal caregiving)||16,475||7632/7632|
|Mortality, Patient, Time, and Other Costsc||Description||Base Case Value, 2010 US $||Range for Sensitivity Analysis, $||Reference(s)|
|Breast cancer mortality||b||Mandelblatt 200539|
|Nonbreast cancer mortality||45,878||b||Levinsky 200141|
|Patient time costs||—|
|Screening mammogram||60||47-86||Secker-Walker 199942|
|MRI||77||55-121||Bureau of Labor|
|Diagnostic mammogram||139||111-193||Statistics 200643|
|Patient time costs (continued)|
|Excisional biopsy||568||523-658||Secker-Walker 1999,42 Bureau of Labor Statistics 200643|
|Breast cancer, clinical presentation||583||502-744|
|Surgery and radiation therapy||3564||b|
|Breast cancer follow-up visits||241||b|
|Wage loss in breast cancer patients, first y after diagnosis||6927||b||Lauzier 200744|
|Informal caregiving costs, first y after cancer diagnosis||609||b||Bureau of Labor Statistics 2006,43 Hayman 200145|
|Other annual costs||—|
|Well, no cancer: Age, y|
Primary outcomes were lifetime costs, quality-adjusted life-years (QALYs), and incremental cost-effectiveness (expressed as dollars per QALY gained). Projected QALYs included 25 years of full health accrued at model entry. Both costs and QALYs were discounted at 3% annually.
Additional outcomes included total undiscounted life expectancy without quality adjustment, breast cancer mortality risk reduction, and potential harms of screening (false-positive screening results, false-positive biopsies, and radiation-induced breast cancer). Component cost analysis was performed by separating lifetime costs into categories related to the direct medical costs of screening and diagnosis, patient time costs for screening and diagnosis, breast cancer treatment (including patient time costs for treatment), breast cancer mortality, and other costs (including direct medical costs unrelated to breast cancer, and nonbreast cancer mortality costs).
We used standard cost-effectiveness analytic methods.47 Incremental cost-effectiveness ratios (ICERs) for a strategy compared with the next most effective strategy were calculated by dividing the difference in costs by the difference in effectiveness to obtain the cost required to gain an additional QALY.
We evaluated the effect of uncertainty on model results using multiple sensitivity analyses (Tables 1 and 2). To characterize the random variability in individual outcomes, we analyzed the model as a Markov Monte Carlo simulation. We then performed univariate sensitivity analyses to examine the effect of additional uncertainty regarding input parameter values on model results. We also conducted 2-way sensitivity analyses of diagnostic test performance using additional published reports of test performance for both mammography16 and MRI.18-21 Multiparameter sensitivity analyses were performed for alternate sets of good-fitting natural history input parameters that were identified during model calibration.14 Prophylactic oophorectomy at ages 35 years, 40 years, and 45 years also was examined, and subsequent breast cancer risk was reduced by half.48, 49 In addition, short-term quality-of-life decreases related to false-positive screening results were included13, 50 (base case: no short-term disutilities from screening), and patient time costs were examined by varying the time lost from work between 50% and 200% of base-case values.42, 43
Lifetime costs, QALYs, and ICERs are listed in Table 3. For women with BRCA1 mutations, adding alternating MRI to annual DM beginning at age 30 years (Alt30) increased QALYs and costs, with an ICER of $74,200 per QALY gained. The MRI25/Alt30 and Alt25 strategies had minimal additional QALY gains and increased lifetime costs with substantially higher ICERs.
|Screening Strategy||Lifetime Costs, $a||QALYsa||ICER, $|
For women with BRCA2 mutations, Alt30 screening had a considerably higher ICER compared with DM30 ($215,700 per QALY gained) and also was the least costly dual-modality screening strategy evaluated. MRI25/Alt30 screening provided minimal additional QALYs and increased lifetime costs. Alt25 was eliminated as a screening strategy, because it provided no additional QALYs and increased lifetime costs relative to MRI25/DM30.
Because Alt30 screening was the least costly approach for implementing dual-modality screening guidelines in both BRCA1 and BRCA2 mutation carriers, we focused on this strategy for subsequent analyses on additional outcomes, component costs, and sensitivity analyses.
Alt30 screening reduced breast cancer mortality compared with DM alone (Table 4). False-positive screening results and false-positive biopsies increased substantially with Alt30 screening compared with DM30 screening alone and occurred more frequently in BRCA2 carriers. Radiation-induced breast cancers represented <2% of all diagnosed breast cancers and remained stable when MRI screening was added, because MRI does not use ionizing radiation.
|Long-Term Outcomes||Potential Harms|
|Screening Strategy||Total LE, y||BC Mortality per 1000 Women DX With BC||% BC Mortality Reduction||No. of FP Screens, Average per Woman||No. of FP Biopsies, Average per Woman||% Radiation- Induced BC|
Component Cost Analysis
Lifetime and component costs increased with dual-modality breast cancer screening (Table 5). The increase in total lifetime costs associated with Alt30 screening was driven primarily by increased screening and diagnostic costs. Although breast cancer treatment and mortality costs decreased in the setting of dual-modality screening, these downstream cost savings only partially offset the increased costs of earlier and more intensive screening. Patient time costs for screening and diagnosis more than doubled when alternating MRI was added to DM30 screening.
|Component Cost Categories, $|
|Direct Medical Costs|
|Screening Strategy||Screening and Diagnosis||Breast Cancer Treatment||Breast Cancer Mortality||Patient Time Costsb||Otherc||Total Lifetime Costs, $|
Univariate and multiparameter sensitivity analyses indicated that, in BRCA1 mutation carriers (Fig. 2A), the most influential parameters for the ICER of Alt30 screening compared with DM30 screening were related to breast cancer risk and MRI cost. As the breast cancer risk decreased, either from a decreased cumulative probability of breast cancer or from prophylactic oophorectomy, dual-modality screening became less cost-effective (increasing the ICER). Varying MRI cost from 50% to 200% of the base-case value of $619 caused the ICER to range from approximately $44,000 to $134,000 per QALY gained. When examining additional reports of MRI test performance, dual-modality screening became more cost-effective if MRI performance at the highest published values19-21 could be achieved, with ICERs <$50,000 per QALY gained.
For BRCA2 mutation carriers (Fig. 2B), the ICER for Alt30 screening compared with DM30 screening similarly was influenced by breast cancer risk and MRI cost, with the ICER increasing substantially with decreasing breast cancer risk. The inclusion of short-term quality-of-life decreases related to false-positive screening results further decreased the cost-effectiveness of Alt30 screening, likely related to the substantially greater frequency of false-positive results in these women.
This analysis evaluated the health outcomes and associated costs of 3 dual-modality screening strategies in women with BRCA gene mutations. Because a prior comparative-effectiveness study7 indicated that dual-modality screening at alternating 6-month intervals would be more effective than combined screening on the same day, we focused our analysis on the cost-effectiveness of alternating strategies only. Our results indicate that alternating DM and MRI screening starting at age 30 years is clinically effective, providing life-expectancy gains and breast cancer mortality reduction, and is also the least costly approach for implementing current screening guidelines.
Although the cost-effectiveness of screening BRCA mutation carriers with annual combined film mammography and MRI has been studied previously,13, 51-53 our analysis included the evaluation of DM as well as the assessment of alternating dual-modality regimens, which reflect more contemporary clinical practice.6 Our model estimated the frequency of radiation-induced breast cancers and false-positive test results in addition to projecting long-term health benefits to more fully determine the consequences of more intensive screening. Our results estimate that radiation exposure accounts for a very small proportion of detected breast cancers, suggesting that the benefits of screening outweigh the concern over this issue for women carrying high-risk BRCA mutations. In addition, a recent study of 1993 women with BRCA mutations demonstrated that exposure to diagnostic radiation before age 30 years was associated with an increased risk of breast cancer (hazard ratio, 1.90; 95% confidence interval, 1.20-3.00),54 supporting our findings that there is limited benefit from mammographic screening before age 30 years.
To date, no randomized controlled trials focused on the long-term outcomes of dual-modality screening in high-risk women have been conducted, making it difficult to validate our results with published studies. A recent study of 594 BRCA mutation carriers indicated that contemporaneous annual screening with mammography and MRI resulted in an overall survival rate at 6 years of 92.7% (95% confidence interval, 79.0%-97.6%).39 When this strategy was simulated by our model in a cohort of the same mean age, the overall survival rate at 6 years was 95.1%.
Dual-modality screening is more cost-effective in BRCA1 carriers than in BRCA2 carriers, primarily because of higher breast cancer incidence in BRCA1 carriers. Sensitivity analyses indicated that, as breast cancer risk increased, dual-modality screening became more cost-effective. Decreasing MRI costs also improved the cost-effectiveness of screening. Short-term quality-of-life decreases from false-positive testing decreased cost-effectiveness in BRCA2 carriers but had little effect on the cost-effectiveness of screening BRCA1 carriers. This can be attributed to the lower breast cancer risk among BRCA2 carriers compared with BRCA1 carriers and, thus, a greater number of false-positive examinations over a lifetime horizon. Although our current results indicate that alternating dual-modality screening beginning at age 30 years is the most cost-effective approach to implementing current guidelines for women with either BRCA1 or BRCA2 mutations, our results also suggest that alternative screening strategies that were not considered in this analysis may be more cost-effective for BRCA2 carriers, and this is an important area for further study. It is also important to note that, if MRI test performance as high as the most optimistic reports in the literature can be achieved, then MRI screening would be more cost-effective than projected in our base-case analysis.
Understanding the impact of screening on component cost categories is critical as we increasingly focus on providing high-value, cost-conscious health care.10, 11, 56 Our evaluation of lifetime and component costs related to dual-modality screening revealed that the increased costs of additional MRI screening exceeded the downstream reduction in breast cancer treatment and mortality costs. This is caused in part by screening costs that apply to all women and accrue in the nearer future, whereas breast cancer treatment and mortality costs occur in fewer women and accrue many years later. Patient time costs for screening and diagnosis more than doubled when MRI screening was added.
Simplifying assumptions of this model included the use of some parameter values from studies of breast cancer in the general population.32, 33, 35 Although we would have preferred to use BRCA carrier-specific parameters throughout the model, the use of general population input values occurred only when BRCA-specific data were sparse or unavailable. Because the magnitude of any differences, if present, is not known, it is difficult to anticipate their effects on the model, and we recognize that model results may change if additional information becomes available. We focused on detecting the first primary breast cancer and assumed perfect adherence to screening and treatment protocols. Medicare reimbursement was used to determine costs of screening and diagnosis, although most of the simulated screening events occurred among women aged <65 years, because Medicare reimbursement is a generalizable and transparent proxy for US health care costs. Finally, although Alt30 screening appears to be a cost-effective dual-modality screening approach for BRCA mutation carriers, women from families in which breast cancer has been diagnosed before age 35 years may benefit from starting screening before age 30 years.
In conclusion, this analysis suggests that screening with MRI and DM at alternating 6-month intervals beginning at age 30 years is a clinically effective approach to applying current clinical guidelines and is considerably more cost-effective in BRCA1 gene mutation carriers compared with BRCA2 gene mutation carriers.
This work was supported by National Institutes of Health (NIH) grants NIH K07-CA128816 (Dr. Lee), NIH K25-CA133141 (Dr. Kong), and NIH R00-CA126147 (Dr. McMahon) and by the Harvard Medical School Office for Enrichment Programming (Dr. Lowry). Provision of selected model input parameter values by the Breast Cancer Surveillance Consortium (BCSC) was supported by the National Cancer Institute (U01CA63740, U01CA86076, U01CA86082, U01CA63736, U01CA70013, U01CA69976, U01CA63731, U01CA70040, and HHSN261201100031C).
CONFLICT OF INTEREST DISCLOSURES
Dr. Gazelle has served as a consultant for GE Healthcare.
- 2National Comprehensive Cancer Network (NCCN). NCCN Clinical Practice Guidelines in Oncology: Breast Cancer. Version 2. Jenkintown, PA: National Comprehensive Cancer Network; 2011.
- 3Breast cancer screening with imaging: recommendations from the Society of Breast Imaging and the ACR on the use of mammography, breast MRI, breast ultrasound, and other technologies for the detection of clinically occult breast cancer. J Am Coll Radiol. 2010; 7: 18-27., , , et al.
- 10High-value, cost-conscious health care: concepts for clinicians to evaluate the benefits, harms, and costs of medical interventions. Ann Intern Med. 2011; 154: 174-180., , , et al.
- 11Appropriate use of screening and diagnostic tests to foster high-value, cost-conscious care. Ann Intern Med. 2012; 156: 147-149., , , et al.
- 30Estrogen receptor expression as a predictive marker of the effectiveness of tamoxifen in the treatment of DCIS: findings from NSABP Protocol B-24 [abstract]. Paper presented at: San Antonio Breast Cancer Symposium, December 11-14, 2002; San Antonio, TX., , , et al.
- 32Breast Cancer Surveillance Consortium. Frequency of node-positivity and metastatic cancer by primary tumor size and method of detection [deidentified data obtained directly from BCSC, September 18, 2006]. Website: http://www.breastscreening.cancer.gov/.
- 34United States life tables, 2006. Natl Vital Stat Rep. 2010; 58: 1-40..
- 35SEER Cancer Statistics Review, 1975-2006 [based on the November 2008 SEER data submission, available online 2009]. Bethesda, MD: National Cancer Institute; 2009., , , et al. eds.
- 36Department of Health and Human Services; Centers for Medicare and Medicaid Services. Medicare Physician Fee Schedule. Available at: http://www.cms.gov/apps/physician-fee-schedule/overview.aspx. Accessed September 14, 2010.
- 37US Department of Labor; Bureau of Labor Statistics. Consumer Price Index. Available at: http://www.bls.gov/cpi/. Accessed October 21, 2010.
- 40Red Book. Montvale, NJ: Thomson Healthcare; 2007.
- 43US Department of Labor; Bureau of Labor Statistics. National Compensation Survey, 2006. Available at: http://www.bls.gov/ncs/data.htm. Accessed on October 21, 2010.
- 47Cost-Effectiveness in Health and Medicine. New York: Oxford University Press; 1996., , , et al. eds.
- 56American College of Physicians Ethics Manual: 6th ed. Ann Intern Med. 2012; 156: 73-104..