Search

We compare and demonstrate the effectiveness of two clustering methods with the main purpose of identifying characteristic profiles of high utilizers of health care. In this work, we use three sets of mutually independent longitudinal data that are nationally representative of the US adult working-age civilian non-institutionalized population. We compare k-means, a commonly used clustering method, with a k-medoids algorithm called Partitioning Around Medoids. We use one cohort of data to create clusters based on similar characteristics of individuals for both clustering methods. We examine these characteristic compositions of the highest three average total expenditure clusters from this cohort. We also examine the health expenditure distributions for this cohort over the following two years. We validate the approach by applying the centers of the clusters to two other cohorts of similar data. We form clusters based on demographic, economic, and health-related characteristics that are commonly used in studies of health care utilization. We demonstrate the consistency of our results across the three cohorts of data and across different types of health expenditures, such as office-based/outpatient and drug. Clusters can be formed with other more homogeneous data, such as Medicaid, Medicare, employer sponsored insurance, or individual private plans issued under the Affordable Care Act. This approach can be used to follow similar groups over time for other types of health outcomes.

This study examines the causal effect of delayed oral health care on increased emergency department visits in the United States. We extend prior research by estimating the effect of delayed or forg...

High utilizers of health care are those individuals in the upper tail of the distribution each year. The main purpose of our work is to identify determinants of persistent high utilizers in adults,...

Background Assessment of drug costs for cost-effectiveness analyses (CEAs) in the United States is not straightforward because the prices paid for drugs are not publicly available and differ between payers. CEAs have relied on list prices that do not reflect the rebates and discounts known to be associated with these purchases. Objectives To review available cost measures and propose a novel strategy that is transparent, consistent, and applicable to all CEAs taking a US health care sector perspective or a societal payer's perspective. Methods We propose using the National Average Drug Acquisition Cost (NADAC), the Veterans Affairs Federal Supply Schedule (VAFSS), and their midpoint as the upper bound, lower bound, and base case, respectively, to estimate net drug prices for various payers. We compare this approach with wholesale acquisition cost (WAC), the most common measure observed in our literature review. The minimum WAC is used to provide the most conservative comparison. Results Our sample consists of 1436 brand drugs and 1599 generic drugs. On average, the upper bound (NADAC) is 1% and 9.8% lower than the WAC for brand and generic drugs respectively, whereas the lower bound (VAFSS) is 48.3% and 54.2% lower than the WAC. The NADAC is less than the WAC in 89.6% of drug groups. The distributions of these relationships do not show a clear mode and have wide variation. Conclusions Our study suggests that the WAC may be an overestimate for the base case because the minimum WAC is higher than the NADAC for most drugs. Our approach balances uncertainty and lack of data for the cost of pharmaceuticals with the need for a transparent and consistent approach for valid CEAs.

SummaryBackground Previous estimates of the cost of care for pediatric Cystic fibrosis (CF) showed wide variation, without specific summary of pulmonary drug costs. Methods Enrolled CF children from the Wisconsin newborn screening trial were evaluated quarterly per protocol. Assessments systematically included all treatments, hospitalizations, and nutritional and pulmonary outcomes. Direct medical costs from hospital billing and medical records from 1989 to 2010 were used to describe costs by age-ranges and subgroups throughout follow-up. Outpatient drugs were separated by category (pulmonary/otherwise). Inpatient and drug costs were examined by clinical risk factors (presence of meconium ileus, pancreatic insufficiency, and expected severity of genetic mutations). Results Seventy-three children were followed for an average of 12.9 years with an average annual total cost of care of $24,768. Outpatient drug costs (53%) and hospitalizations (32%) represented the majority of costs. Drug costs were 48% for pulmonary indications and 52% for non-pulmonary. Pulmonary drug costs for children taking dornase were 54% of their drug costs while pulmonary drug costs were only 31% for children not taking dornase. Significant differences in frequency of inpatient stays existed for children with pancreatic insufficiency. Substantial differences in treatment costs exist as children age and by clinical risk factor. Conclusion This study provides more accurate longitudinal estimates of CF care costs throughout childhood and shows that increasing age, pancreatic insufficiency, use of dornase, and hospitalizations are key determinants of cost. These estimates can be included in evaluations of the cost-effectiveness of new, highly expensive treatments being introduced for any CF population. Pediatr Pulmonol. 2016; 9999:1–9. © 2016 Wiley Periodicals, Inc.

Black women tend to be diagnosed with breast cancer at a more advanced stage than whites and subsequently experience elevated breast cancer mortality. We sought to determine whether there are racial differences in tumor natural history that contribute to these disparities. We used the University of Wisconsin Breast Cancer Simulation Model, a validated member of the National Cancer Institute's Cancer Intervention and Surveillance Modeling Network, to evaluate the contribution of racial differences in tumor natural history to observed disparities in breast cancer incidence. We fit eight natural history parameters in race-specific models by calibrating to the observed race- and stage-specific 1975-2000 U.S. incidence rates, while accounting for known racial variation in population structure, underlying risk of breast cancer, screening mammography utilization, and mortality from other causes. The best fit models indicated that a number of natural history parameters must vary between blacks and whites to reproduce the observed stage-specific incidence patterns. The mean of the tumor growth rate parameter was 63.6 % higher for blacks than whites (0.18, SE 0.04 vs. 0.11, SE 0.02). The fraction of tumors considered highly aggressive based on their tendency to metastasize at a small size was 2.2 times greater among blacks than whites (0.41, SE 0.009 vs. 0.019, SE 0.008). Based on our simulation model, breast tumors in blacks grow faster and are more likely to metastasize earlier than tumors in whites. These differences suggest that targeted prevention and detection strategies that go beyond equalizing access to mammography may be needed to eliminate breast cancer disparities.

OBJECTIVE:Because cystic fibrosis can be difficult to diagnose and treat early, newborn screening programs have rapidly developed nationwide but methods vary widely. We therefore investigated the costs and consequences or specific outcomes of the 2 most commonly used methods.METHODS:With available data on screening and follow-up, we used a simulation approach with decision trees to compare immunoreactive trypsinogen (IRT) screening followed by a second IRT test against an IRT/DNA analysis. By using a Monte Carlo simulation program, variation in the model parameters for counts at various nodes of the decision trees, as well as for costs, are included and applied to fictional cohorts of 100 000 newborns. The outcome measures included the numbers of newborns given a diagnosis of cystic fibrosis and costs of screening strategy at each branch and cost per newborn.RESULTS:Simulations revealed a substantial number of potential missed diagnoses for the IRT/IRT system versus IRT/DNA. Although the IRT/IRT strategy with commonly used cutoff values offers an average overall cost savings of $2.30 per newborn, a breakdown of costs by societal segments demonstrated higher out-of-pocket costs for families. Two potential system failures causing delayed diagnoses were identified relating to the screening protocols and the follow-up system.CONCLUSIONS:The IRT/IRT screening algorithm reduces the costs to laboratories and insurance companies but has more system failures. IRT/DNA offers other advantages, including fewer delayed diagnoses and lower out-of-pocket costs to families.

This article extends the standard two-part model for predicting health care expenditures to the case where multiple events may occur within a one-year period. The first part of the extended model represents the frequency of events, such as the number of inpatient hospital stays or outpatient visits, and the second part models expenditure per event. Both component models also use independent variables that consist of an individual’s demographic and access characteristics, socioeconomic status, health status, health insurance coverage, employment status, and industry classification. The second part of the model also includes a variable representing the number of events to predict the expenditure per event, thus capturing dependencies between the first and second parts. This article introduces closed-form predictors of annual total expenditures and demonstrates how to create simulated predictive distributions for individuals and groups. The data for this study are from the Medical Expenditure Panel ...

Health care expenditures have accounted for increasing proportions of the U.S. gross domestic product, and the rate of growth of health care expenditures has increased over the past two decades. These two measures of assessing whether the level of health care expenditures is affordable may be appropriate in the aggregate for the United States but are not appropriate to assess whether individual stakeholder groups can afford their particular level of spending on health care. Health care is an economic good that differs from other economic goods, as it involves life and death issues, and invokes a call for a moral authority. This article explores definitions of what is affordable health care from the perspective of different stakeholders and suggests that other measures are needed to assess whether or not health care is affordable for stakeholders as one definition is not appropriate for all stakeholders. INTRODUCTION Financial forecasts for U.S. health care spending have led to concerns as to whether the economy can sustain this growth in the future, or whether there is a potential crisis in the level of health care spending today. Two common measures used to indicate that the U.S. health care spending is not affordable with regard to the level of aggregate spending are: a growing percentage of gross domestic product (GDP) devoted to health care and increasing rates of growth of health care spending. What is overlooked in these measures is whether they are applicable to different stakeholder groups or stakeholders at an individual level and what financial impact they may have on stakeholders. Stakeholders are health care providers, private payers such as insurers, public payers such as governmental entities, employers, and consumers who utilize the services, who also may pay some or all of the premiums, and may also pay the providers. This article will address these stakeholder perspectives and suggest that their perspectives are critical to the discussion of whether or not health care is affordable. As background, in 1960, U.S. National Health Expenditures (NHE) were 5.2 percent of the U.S. GDP, while in 2005, NHE were 16 percent of GDP (Centers for Medicare & Medicaid Services, 2007). The Office of the Actuary at the Centers for Medicare and Medicaid Services (CMS) developed macro-level models for projecting short-term and long-term health care spending for the U.S. economy (Braden et al., 1998; Smith et al., 1999; Seiden et al., 2001). Short-term national-level forecasts from CMS estimate growth in NHE to $3.6 trillion in 2014, or 18.7 percent of GDP (Heffler et al., 2005). By 2014, the public sector is estimated to account for half of total health care spending (Centers for Medicare & Medicaid Services, 2007). Intermediate-term projections using the CMS methodology indicated that NHE would increase to $16.0 trillion by 2030, or 32 percent of GDP. The average nominal annual growth rate of NHE was projected as 8.3 percent from 1990 to 2030, while the projected real annual NHE growth rate was about 3.0 percent (Burner et al., 1992). Chernew et al. (2003) suggested two alternative definitions for health care affordability in the aggregate. The first definition examined the difference between national income and a level of nonhealth care spending. If the difference was positive, then health care spending was deemed to be affordable. A second definition, admittedly more conservative, examined whether a proportion of the increase in income could be spent on health care. Either definition would allow the absolute amount of money spent on health care to rise with income. Their article utilized the second definition to be consistent with a Medicare Technical Review panel that assumed there would never be a downward trend in nonhealth care spending. Their conclusions were that aggregate health care expenditures would be affordable until 2075 if real annual NHE grew at a rate 1 percentage point higher than real annual GDP, even though aggregate health care expenditures would consume 38 percent of GDP. …

Rationale and Objectives The aims of this study were to determine diagnostic radiology faculty members' compliance with recommended health guidelines for physical activity, body weight, diet, and related health indicators and to compare faculty members' compliance with that of radiology residents. Materials and Methods A request to complete an online health survey was electronically sent to members of the Association of University Radiologists in September 2008. Results were compared to those from a similar survey completed by radiology residents in May and June 2007. Frequency counts and Fisher's exact tests were used to summarize results and to determine statistically significant relationships. Results The sample consisted of 193 of 801 members of the Association of University Radiologists (24%). A greater percentage of faculty members than residents complied with recommendations for physical activity (52% vs 37%, P P P P P = .002) and did not think about stress on most days (39% vs 26%, P = .001). Most faculty members (59%) worked 51 to 60 hours a week, whereas most residents (59%) worked >60 hours, and greater work hours were correlated with less resident physical activity ( P = .017). More female than male faculty members (78% vs 57%, P = .010) and residents (83% vs 62%, P 2 . Conclusion A substantial percentage of faculty members were out of compliance with federal health guidelines, although less so than residents in many categories. Comments from both groups suggest a possible benefit from modifications to the work environment.

Chronic diseases account for 75% of U.S. national health care expenditures as estimated by the Centers for Disease Control. Many chronic diseases are punctuated by acute episodes of illnesses that occur randomly and create cost spikes in utilization from one year to the next. Modeling to account for these random events provides better estimates of (1) future costs and (2) their variability. A Bayesian statistical model is used to predict the incidence and cost of hospitalizations for one chronic disease. A two-part statistical model is described that separately models the utilization and cost of hospitalization. Individual demographic characteristics are included as well as a simple biological classification system to adjust for the severity of disease among individuals. Results by child, as well as by calendar year, are presented. Using a simple approach to incorporate severity, the model produces reasonable estimates of the number of hospitalizations and cost of hospitalization for the group in...

Background: Simulation models analyzing the impact of treatment interventions and screening on the level of breast cancer mortality require an input of mortality from causes other than breast cancer, or competing risks. Methods: This chapter presents an actuarial method of creating cohort life tables using published data that removes breast cancer as a cause of death. Results: Mortality from causes other than breast cancer as a percentage of all-cause mortality is smallest for women in their forties and fifties, as small as 85% of the all-cause rate, although the level and percentage of the impact varies by birth cohort. Conclusion: This method produces life tables by birth cohort and by age that are easily included as a common input by the various CISNET modeling groups to predict mortality from other causes. Attention to removing breast cancer mortality from all-cause mortality is worthwhile, because breast cancer mortality can be as high as 15% at some ages.

The Wisconsin Breast Cancer Epidemiology Simulation Model is a discrete-event, stochastic simulation model using a systems-science modeling approach to replicate breast cancer incidence and mortality in the U.S. population from 1975 to 2000. Four interacting processes are modeled over time: (1) natural history of breast cancer, (2) breast cancer detection, (3) breast cancer treatment, and (4) competing cause mortality. These components form a complex interacting system simulating the lives of 2.95 million women (approximately 1/50 the U.S. population) from 1950 to 2000 in 6-month cycles. After a "burn in" of 25 years to stabilize prevalent occult cancers, the model outputs age-specific incidence rates by stage and age-specific mortality rates from 1975 to 2000. The model simulates occult as well as detected disease at the individual level and can be used to address "What if?" questions about effectiveness of screening and treatment protocols, as well as to estimate benefits to women of specific ages and screening histories.

Background: Many guidelines recommend screening mammography every 1 – 2 years for women older than 40 years; more than 70% of women now participate in routine screening. No studies have examined the societal impact of screening practices over the past decade in the United States on costs and quality-adjusted life-years (QALYs). We performed a re trospective cost-effectiveness analysis comparing actual and alternative screening mammography scenarios. Methods: We used a discrete-event simulation model of breast cancer epidemiology to estimate the costs and the number of QALYs that were associated with observed screening mammography patterns in the United States from 1990 to 2000 for women aged 40 years or older. We also estimated costs and QALYS for no screening and for 64 alternative screening scenarios. Incremental cost-effectiveness ratios were computed. Sensitivity analyses were performed on key parameters. Results: Actual U.S. screening patterns from 1990 to 2000 accrued 947.5 million QALYs and cost $166 billion over the lifetimes of the screened women, resulting in a gain of 1.7 million QALYs for an additional cost of $62.5 billion compared with no screening. Among those polices that were not dominated — i.e., for which no alternative existed that produced more QALYs for lower costs — screening all women aged 40 – 80 years annually per some U.S. guidelines was the most expensive option, costing $58 000 per additional QALY gained compared with the next most costly alternative, screening all women aged 45 – 80 years annually. Many alternative screening scenarios generated more QALYs for less cost (with savings up to $6 billion) than actual screening patterns over the study period. Sensitivity analysis showed that conclusions about the costeffectiveness of screening mammography policies were highly sensitive to small, short-term detrimental effects on quality of life from the screening test itself. Conclusions: Choosing among the effi cient policies to guide current screening recommendations requires consideration of costs to promote participation in screening and measurement of acute quality-of-life effects of mammography. [J Natl Cancer Inst 2006;98:774 –82 ]

Background Although there are more than 1000 mutations in the cystic fibrosis transmembrane conductance regulator (CFTR) gene, most of them are uncommon and only limited information exists regarding genotype–pulmonary phenotype relationships. Methods We determined and classified the CFTR mutations using denaturing high-performance liquid chromatography and developed new, quantitative methods to categorize pulmonary phenotypes. Results Two novel alleles were discovered, namely G1047R and 1525-2A→G, which were accompanied by F508del and G551D mutations, respectively. Assessment of numerous options revealed that CF pulmonary phenotype categorization in children cannot be accomplished with clinical or pulmonary function data but is facilitated by longitudinal quantitative chest radiology. It was most useful to categorize pulmonary disease status by evaluating the typical pattern of abnormalities in patients homozygous for the F508del mutation, and then compare patients with minor mutations to this typical CF pulmonary phenotype. By this method, both patients with novel mutations have pulmonary phenotypes typical of F508del homozygotes. However, patients with class IV mutations (e.g., R347P) or with pancreatic sufficiency showed serial chest radiographs that were atypically mild. Conclusions Longitudinal quantitative chest radiography provides a new strategy for CF pulmonary phenotype categorization that should be useful for genotype–phenotype delineation in individual patients and in both epidemiologic studies and clinical trials involving groups of children with CF.

Objectives To extend previous evaluations of costs of cystic fibrosis (CF) diagnosis and examine key issues in assessing the CF cost of care. Study design Costs for CF newborn screening (NBS) including CF multi-mutation testing are analyzed by using data from the Wisconsin State Laboratory of Hygiene. Electronic data from 2 Wisconsin CF centers are used to illustrate the complexity of analyzing CF health care utilization and costs. Results The current cost-per-newborn of a CF multi-mutation test is 50% higher than testing for a single mutation. Data collection for the cost-of-care study requires a combination of electronic and manual data collection; modeling of cost data requires consideration of any censoring. Hospitalizations are shown to have a large impact on costs and show high variability at the individual level. Sixty-nine percent of children with meconium ileus had some hospitalization versus 56% of children without meconium ileus. Conclusion A cost-benefit analysis of CF multi-mutation testing is warranted. The study of health care cost data is complex and utilization varies between children. Individual-level modeling of CF costs must include factors contributing to the severity of the disease and allow for consideration of individual-level utilization, such as the number of hospitalizations.

A health care measure that assesses the occurrence or non-occurrence of an event may be costly to obtain. For instance, measuring whether an inpatient hospital stay was necessary according to accepted clinical protocols is a relevant, though expensive analysis. Prior research (Rosenberg, 1998; Rosenberg, et al., 1999; Rosenberg and Griffith, 2000) has shown that a Bayesian logistic regression model is an excellent tool to predict the rate of Non-Acceptable inpatient hospital Claims (NACs). This paper describes a decision-theoretic approach to determine whether a change in the NAC rate warrants investigation. The Bayesian model is used to estimate, by principal discharge diagnosis, the posterior distribution of the NAC rate for a group of claims at the time of the audit (defined as time 0). This distribution of the NAC rate is compared against another posterior distribution of the NAC rate obtained at a later time (without further audit). Claim and investigation costs are included in the assessment of whether an investigation of a change in the NAC rate should be undertaken. This approach provides an inexpensive, supplemental level of control between audits for monitoring the level of the NAC rate. The methodology is also applicable to other health care measurements.

The Medicare program, private insurers, and managed care organizations reimburse hospitals for inpatient admissions using the Diagnosis Related Group (DRG). The DRG is determined from a complicated algorithm based on patient medical records. Previous studies generated concerns about ‘DRG upcoding’, in which incorrect DRG codes may be selected by the hospital to obtain higher reimbursement. Insurers rely on expensive manual audits of claims to verify the appropriateness of the DRG coding. A statistical system that can adaptively detect claims with incorrect DRG codes would provide a powerful improvement to current practice. This paper describes two aspects of the statistical system that provides proof that the concept is viable. The first aspect of the paper is the design of a hierarchical Bayesian model to be applied to claims data (without audit) to estimate the probability that a claim is coded incorrectly. The second aspect of the paper is the use of the Bayesian model to aid in the selection process of claims to audit by proposing that a claim should be investigated if the predicted recovery is more than the cost of auditing that claim. This approach improves upon that used currently by auditing 88% of the claims and recovering 98% of the overpayments. While these results improve upon the current approach for determining which claims to investigate, they are based on data that have been systematically selected for audit based on one insurer's past experience. Future work will create an adaptive system to determine the selection of claims to audit from the entire paid claims database, and that can be generalized for use by other insurers.

Background. Simulation models designed to evaluate cancer prevention strategies make assumptions on background mortality—the competing risk of death from causes other than the cancer being studied. Researchers often use the U.S. life tables and assume homogeneous other-cause mortality rates. However, this can lead to bias because common risk factors such as smoking and obesity also predispose individuals for deaths from other causes such as cardiovascular disease. Methods. We obtained calendar year-, age-, and sex-specific other-cause mortality rates by removing deaths due to a specific cancer from U.S. all-cause life tables. Prevalence across 12 risk factor groups (3 smoking [never, past, and current smoker] and 4 body mass index [BMI] categories [2]) were estimated from national surveys (National Health and Nutrition Examination Surveys [NHANES] 1971–2004). Using NHANES linked mortality data, we estimated hazard ratios for death by BMI/smoking using a Poisson regression model. Finally, we combined these results to create 12 sets of BMI and smoking-specific other-cause life tables for U.S. adults aged 40 years and older that can be used in simulation models of lung, colorectal, or breast cancer. Results. We found substantial differences in background mortality when accounting for BMI and smoking. Ignoring the heterogeneity in background mortality in cancer simulation models can lead to underestimation of competing risk of deaths for higher-risk individuals (e.g., male, 60-year old, white obese smokers) by as high as 45%. Conclusion. Not properly accounting for competing risks of death may introduce bias when using simulation modeling to evaluate population health strategies for prevention, screening, or treatment. Further research is warranted on how these biases may affect cancer-screening strategies targeted at high-risk individuals.

Objectives To compare the cost of diagnosing cystic fibrosis (CF) through a newborn screening program with the traditional method and to estimate the cost of CF diagnosis if a national newborn screening program is implemented. Study design Surveys were conducted to determine the annual number of sweat tests in 1991 and in 2000 after implementation of statewide screening. A national survey of sweat test costs was used to estimate the annual expense for diagnosing CF in the United States through newborn screening. Results Since the introduction of newborn screening for CF, the numbers of sweat tests ordered annually have decreased from 1670 to 804 (including 134 follow-up tests from screening). The current estimated annual cost of Wisconsin CF newborn screening and diagnosis is $4.58 per newborn infant. The estimated annual cost per newly diagnosed CF infant using the traditional method is $4.97 per newborn infant. If no additional sweat tests were ordered outside of the newborn screening program, the estimated annual cost of a Wisconsin CF newborn screening and diagnosis is $2.66 per newborn and $2.47 per newborn for a national CF newborn screening program. Conclusions A CF newborn screening program provides a potentially cost-saving alternative to the traditional method of diagnosis of CF. (J Pediatr 2003;142:617-23)

OBJECTIVES: To ascertain the variation in strategies for managing delirium of physicians with expertise in geriatrics. DESIGN: Cross-sectional mail survey. SETTING: United States. PARTICIPANTS: A probability sample of physician members of the American Geriatrics Society. MEASUREMENTS: Management choices presented in a two-part case vignette of an older woman hospitalized with a hip fracture who develops mild and then severe delirium. RESULTS: One hundred twenty-two respondents (43%) selected the three answers constituting current “best practice,” 50 (18%) selected an unnecessary diagnostic test (brain imaging, lumbar puncture, or electroencephalogram), and 47 (17%) selected unnecessary pharmacologic therapy for mild delirium. For severe delirium, 270 (96%) selected pharmacological therapy, of whom 180 chose haloperidol alone, 55 chose lorazepam alone, 23 chose lorazepam in combination with haloperidol, and 12 wrote in another drug. Thirty percent of the respondents made any selection of lorazepam, alone or in combination with haloperidol, for mild or severe delirium. Sixty-one percent of those selecting haloperidol for severe delirium chose a dose greater than that recommended for geriatric patients. Sex, date of graduation from medical school, clinical specialty, completion of a geriatric fellowship, or certification in geriatrics had no significant effect on responses. CONCLUSIONS: The common selection of lorazepam to treat delirium is troubling because benzodiazepines themselves are implicated in delirium. Selection of an initial dose of haloperidol higher than that recommended for geriatric patients by more than half of the respondents is also of concern. There is a paucity of sound clinical evidence to guide the choice of pharmacological agents for treating delirium in older hospitalized patients.

Bayesian statistics provide effective techniques for analyzing data and translating the results to inform decision making. This paper provides an elementary tutorial overview of the WinBUGS software for performing Bayesian statistical analysis. Background information on the computational methods used by the software is provided. Two examples drawn from the field of medical decision making are presented to illustrate the features and functionality of the software.

Importance Breast magnetic resonance imaging (MRI) is highly sensitive for detecting breast cancer. Low specificity, cost, and little evidence regarding mortality benefits, however, limit recommendations for its use to high-risk women. How breast MRI is actually used in community settings is unknown. Objective To describe breast MRI trends and indications in a community setting. Design, Setting, and Participants Retrospective cohort study at a not-for-profit health plan and multispecialty group medical practice in New England of 10 518 women aged 20 years and older enrolled in the health plan for at least 1 year who had at least 1 breast MRI between January 1, 2000, and December 31, 2011. Main Outcomes and Measures Breast MRI counts were obtained from claims data. Clinical indication (screening, diagnostic evaluation, staging or treatment, or surveillance) was determined using a prediction model developed from electronic medical records on a subset of participants. Breast cancer risk status was assessed using claims data and, for the subset, also through electronic medical record review. Results Breast MRI use increased more than 20-fold from 6.5 per 10 000 women in 2000 to 130.7 per 10 000 in 2009. Use then declined and stabilized to 104.8 per 10 000 by 2011. Screening and surveillance, rare indications in 2000, together accounted for 57.6% of MRI use by 2011; 30.1% had a claims-documented personal history and 51.7% a family history of breast cancer, whereas 3.5% of women had a documented genetic mutation. In the subset of women with electronic medical records who received screening or surveillance MRIs, only 21.0% had evidence of meeting American Cancer Society (ACS) criteria for breast MRI. Conversely, only 48.4% of women with documented deleterious genetic mutations received breast MRI screening. Conclusions and Relevance Breast MRI use increased steeply over 10 years and then stabilized, especially for screening and surveillance among women with family or personal history of breast cancer; most women receiving screening and surveillance breast MRIs lacked documented evidence of meeting ACS criteria, and many women with mutations were not screened. Efforts are needed to ensure that breast MRI use and documentation are focused on those women who will benefit most.

Purpose: Ductal carcinoma in situ (DCIS), a nonobligate precursor to invasive breast cancer, makes up 20% of new breast cancer diagnoses. DCIS is primarily detected by mammography and has excellent prognosis regardless of therapy choice. We aim to quantify the impact of detection and treatment of DCIS on breast cancer mortality using computer modeling. Methods: We used a validated microsimulation model to replicate U.S. trends in breast cancer incidence and mortality from 1975 to 2000. All breast cancer tumors began in the in situ stage and progressed through more advanced stages with increasing tumor size and lymph node involvement, with a fraction of tumors assumed as nonlethal. Age-adjusted breast cancer mortality was examined under historical U.S. screening and treatment patterns and three alternative scenarios: (1) only women 50 years and older received screening and all breast cancer cases received treatment according to historical patterns, (2) all women received screening according to historical patterns but DCIS was untreated unless later detected at an invasive stage, and (3) historical screening was used but DCIS was never treated. Results: The model predicts breast cancer mortality rates declined during the 1990s to a low of 39.9 per 100,000 in 2000, which is consistent with observed mortality based on the SEER program (38.0 per 100,000 in 2000). Delaying mammography until age 50 or postponing treatment until DCIS progressed to invasive cancer increased mortality to 42.4 and 43.2 per 100,000, respectively, by 2000. If DCIS patients did not receive any treatment, mortality remained steady (∼51.2 per 100,000) through 2000. Conclusions: Our model results suggested detection and treatment of DCIS in the U.S. reduced breast cancer deaths in the year 2000 by about 28%. Most of this benefit (92%–94%) could also be achieved if instead mammography began at age 50 or treatment was reserved for invasive breast cancers only. This abstract is one of the 14 highest scoring abstracts of those submitted for presentation at the 35th Annual Meeting of the American Society of Preventive Oncology, held March 5–8, 2011 in Las Vegas, NV.