The Center undertakes studies across a wide array of domains and methodologies within the healthcare environment.  We have a large array of projects that look at patient care related to care of cancer and skilled nursing/home health patients, access to care for the Medicare and Medicaid populations, as well as developing novel methods for analysis that are at the forefront of statistical research. The methodologies studied provide a theoretical underpinning for understanding which method or sets of methods could be applied to specific issue. Below is a sample of projects utilizing data at the CHDA.  Select the title to learn more.

Methodology

  • Secondary data analysis to predict treatment response based on a large set of covariates using novel ensembling machine learning methods

    Project Lead: Sherri Rose

    Background:

    We will study treatment response in the Sequenced Treatment Alternatives to Relieve Depression (STAR*D) cohort. Specifically, the proposed research is a secondary data analysis to predict treatment response based on a large set of covariates using novel ensembling machine learning methods. Previous work has focused on a limited number of covariates and less flexible estimation methods. The goal is to develop a better understanding of risk factors for treatment-resistant depression by producing a “risk calculator” for clinical use. This risk calculator will allow clinicians to assess the probability a patient has for successful treatment response at presentation.

    Methods Overview:

    Given the size and complexity of clinical data, standard parametric methods may not be suitable. Machine-learning techniques have improved abilities for detecting interaction, nonlinear, and higher-order effects. Machine learning methods aim to “smooth” over the data similarly to parametric regression procedures, but they make fewer assumptions and adapt more flexibly to the data. We will produce a “risk calculator” considering all potential predictors (“full set”) as well as a “risk calculator” with the best small subset of variables (e.g., 10) such that the performance of the smaller set is the closest to the performance of the full set amongst all small sets.

  • Improving Sampling Techniques for Medicare Advantage Plan Payment Methodology with Machine Learning

    Project Lead: Sherri Rose

    Background:

    This pilot study aims to improve risk adjustment for Medicare Advantage plan payment by developing machine learning based matching methods to draw an improved sample of subjects for estimating risk adjustment. The study will then assess the impact of the new sample selection methodology on risk adjustment scores in existing formulas. Risk adjustment models for plan payment are typically estimated using classical linear regression models, and are designed to predict plan spending, often as a function of age, gender, and diagnostic conditions. The use of novel machine learning techniques may improve estimators for risk adjustment, including reducing the ability of insurers to “game” the system with aggressive diagnostic upcoding.

  • Bayesian Methods for Comparative Effectiveness Research with Observational Data

    Project Lead: Sharon-Lise Normand

    Background:

    Health information growth has created unprecedented opportunities to evaluate treatment effectiveness in large and broadly representative patient populations but where the benefits of treatments may vary across population subgroups. We develop novel statistical methods for estimating causal effects that (a) account for uncertainty in the selection of subgroups and for selection of measured confounders; and (b) accommodate unmeasured confounders that moderate treatment effects, in settings where the number of confounders is large and where no randomization has occurred. We illustrate the new methods to answer several substantive questions raised in our ongoing interdisciplinary collaborations that motivate our methods development. To enable reproducible research, we will develop and disseminate SAS macros and R functions.

  • Statistical Methods for Hospital Profiling

    Background:

    We develop a Bayesian framework for semi-continuous competing risk data within both clustered and unclustered settings, with a focus on palliative/End-Of-Life (EOL) care. For EOL care, death is a truncating event and cannot be ignored.  The study of a non-terminal event (e.g. readmission) that is subject to a terminal event (e.g. death) is known as the ‘semi-competing risks’ problem. Current national quality of care assessment efforts ignore the semi-competing risks problem, in large part, because the analysis of semi-competing risks data in the context of clustered observations has not been considered in the literature.  Methods will be illustrated using data on all Medicare enrollees merged with complete tumor data from SEER-Medicare having pancreas, lung, colon, or brain cancer.

  • P01: Data and Methods Core

    Project Lead: Thomas McGuire

    Background:

    The Data & Methods Core provides the personnel, expertise, and computational resources needed for effective use of Medicare data and other data to be acquired and analyzed by the investigators in the Program Project. Additionally, it provides leadership on the data-related issues common to all projects as well as day-to-day direction in the management of complex data sets and analyses. This Core involves four aims. 

    First, the Core is responsible for all data management and oversight activities including dataset acquisition, preparation, integration, management, and quality control. Vanessa Azzone is the Dataset Co-Director.

    Second, the Core offers and coordinates clinical expertise in measurement, specification of key analytic variables, and interpretation across the projects. Almost all aims in the Program Project projects involve some clinical or health data. The explicit clinical linkage will help ensure cross-project learning about fruitful approaches and maximize consistency in approaches across projects. Bruce Landon is the Clinical Co-Director.

    Third, the Core offers and coordinates expertise on statistical model building and other data analysis methods across all projects. A coordinated approach is particularly important given the common features across projects in the data used and interrelated research questions. Alan Zaslavsky is the Statistical Co-Director.

    Fourth, the Core applies theory from health economics to supply- and demand-side setting payment innovations. Thomas McGuire is also the Theory Co-Director.

    The Data & Methods Core is staffed by senior researchers with extensive experience with the data and methods involved in the Program Project projects. 

    View all PO1 Projects here...

Cancer

  • Effectiveness of Chemotherapy for Advanced Ovarian Cancer

    Project Lead: Nancy Keating

    Background:

    Understanding the Effectiveness of Chemotherapy for Advanced Ovarian Cancer to Improve Patient Decisions.  

    Each year, nearly 22,000 women are diagnosed with ovarian cancer, and 62% have advanced disease at diagnosis. Advanced ovarian cancer is not curable, but 70% of women benefit from chemotherapy, at least initially. Few data are available about the benefits versus harms from multiple rounds of different chemotherapies. Patients and physicians need better information to inform decisions about late-line chemotherapy (defined as ≥2 non-platinum-based chemotherapy regimens) that incorporates a patient’s personal characteristics, preferences, and previous response to treatment, to ensure that patients receive treatment that is congruent with their goals.

    This project will address the question, “Given my personal characteristics, conditions and preferences, what should I expect will happen to me?” for women with advanced ovarian cancer who are considering additional chemotherapy as their cancer progresses. Using SEER-Medicare data, we will apply advanced statistical methods to infer the effectiveness and toxicity of chemotherapy for advanced ovarian cancer. We will also examine treatment heterogeneity, with a goal of estimating treatment effectiveness and toxicity for subgroups of women, based on age, existing comorbid illness, and prior treatment response.

  • Factors Associated with Kidney Cancer Diagnosis, Treatment, and Outcomes

    Project Lead: Nancy Keating

    Background:

    We use SEER-Medicare data to assess prognostic factors, treatment, follow-up, and outcomes for a population-based cohort of older patients diagnosed with kidney cancer. Specifically, we will assess differences by patient factors, including race/ethnicity and socioeconomic status and provider factors (including physician and hospital volume) and area-level variations in care. Furthermore, the current literature investigating the use of active surveillance for management of small renal tumors is largely composed of small, single institution, retrospective studies. There are currently no large population-based cohort studies assessing trends in the use of active surveillance for small, incidentally found, renal tumors. Therefore given that there are still no large-scale, prospective data regarding the risks associated with observing small renal lesions, pursuing active surveillance of a renal lesion continues to represent a calculated risk by the treating physician and the affected patient. We want to evaluate the national trends in the use of active surveillance for small, incidentally found, renal tumors and the risks associated with this type of management. Finally, it has been shown that carefully selected patients with good performance status undergoing nephrectomy and subsequent metastasectomy may experience prolonged survival. However, indications for metastasectomy in patients with kidney cancer are not defined. Moreover, there is no clear recommendation for the optimal treatment strategies for patients that have already undergone previous metastasectomy, and were later diagnosed to have recurrent metastasis. We want to further characterize the role of metastasectomy in patients with advanced disease.

  • Treatment Patterns and Outcomes for Women with HER2 Breast Cancer

    Project Lead: Nancy Keating

    Background:

    Breast cancer is the leading cause of cancer deaths in women. Nevertheless, in some subgroups of women, significant improvements in survival time have been achieved. Over the last years, we performed several studies looking at treatment patterns and outcomes among breast cancer patients using population based registries to examine how these advances translated to the overall population care and outcomes. For example, we examined the impact of race on survival of women with metastatic breast cancer by disease subgroup over the last decade and showed that among women with de novo metastatic breast cancer, racial differences in survival were only apparent for those treated with trastuzumab. We also performed several other studies looking at patterns of care and outcomes among patients with early HER2-positive disease. Now, we are planning to expand our studies using one extra year of data, and incorporating both the HER2 status, Part D and Hospice data. Specific Aims include:

    1. Assess endocrine therapy, chemotherapy and trastuzumab utilization rates among women with breast cancers for (a) stage I, II, III and IV breast cancers.
    2. Characterize early and long term overall and breast cancer-specific early among patients with stage I, II, III and IV breast cancers by disease subtype.
    3. Assess patterns of care at the end of life for women with metastatic breast cancer, including hospital admissions.

Medicare

  • Technology Diffusion under New Payment and Delivery Models

    Project Lead: Sharon-Lise Normand
    Technology

    Background:

    This study intends to examine the relationship between organizational traits and diffusion of new medical technology, and to explore the impact of new health care delivery models on costs and quality of health care associated with the new technologies. Study objectives include the comparison of the speeds and mean adoptions times across types of technologies (drugs, devices, and biologics) within disease areas, across lower and higher value technologies, within and across organizations. Our overall objective involves providing quantitative summaries of the associations of new risk-based payment models with fundamental decisions about technology adoption at the organizational level.

    In Aim 1 we will study the diffusion (speed, mean adoption times, ceilings) of selected new technologies as a function of organizational characteristics to determine whether decisions to use new technologies are correlated among different technology types, or within and between disease conditions, within and between organizations.

    In Aim 2, we will distinguish higher from lower value services, identify organizational factors predictive of their use, and determine if organizational decisions to adopt higher vs. lower value services are correlated. Using the findings from Aims 1 and 2,

    Aim 3 will assess the impact of the Medicare risk-based sharing arrangements on total spending, new technology spending, and spending on higher and lower value services, adjusting for organizational factors including organizational-specific diffusion propensities.

  • Risk Adjustment Redesign

    Project Leads: Sherri Rose and Thomas McGuire

    Background:

    This project proposes a transformative redesign of the practice of risk adjustment used for paying health plans in health insurance markets, including Exchanges and Medicare Advantage. Broadly, we intend to compare and propose improvements on existing plan payment methods. The underlying purpose is to simplify and rationalize health plan payment by radically increasing the use of risk-sharing methods and radically reducing the weight on regression based risk adjustment technologies. Our research aims are

    • To characterize the de facto properties of the diagnosis-based risk adjustment payment systems in Exchanges and Medicare in terms of the fit of payments to costs and the incentives for plans to provide the efficient balance of services.
    • To compare the performance of current diagnosis-based risk adjustment systems to alternative payment systems, emphasizing risk-sharing.
    • To improve performance of payment systems by choosing the best combination of tools used to adjust for variation in risk.

    Primary data being used is the MCBS cost and use files 2005-2012.

  • National Implementation of Medicare Advantage and Prescription Drug Plan CAHPS Survey

    Project Lead: Alan Zaslavsky

    Background:

    CAHPS is a patient experience surveys.  These surveys ask patients (or in some cases their families) about their experiences with, and ratings of, their health care providers and plans, including hospitals, home health care agencies, doctors, and health and drug plans, among others. The surveys focus on matters that patients themselves say are important to them and for which patients are the best and/or only source of information.  CMS publicly reports the results of its patient experience surveys, and some surveys affect payments to CMS providers.

    Experience is not the same as Satisfaction
    Patient experience surveys sometimes are mistaken for customer satisfaction surveys. Patient experience surveys focus on how patients experienced or perceived key aspects of their care, not how satisfied they were with their care. Patient experience surveys focus on asking patients whether or how often they experienced critical aspects of health care, including communication with their doctors, understanding their medication instructions, and the coordination of their healthcare needs.  They do not focus on amenities.  

    The Center's focus involves sample design and methodology, case-mix adjustors, construction of data files, sample weights and adjustments. The HMS contribution to the study will involve the use of de-identified data provided by the primary contract, the RAND Corporation.  Work includes sample design, methodology design, case-mix adjustors, linking data files and development of the CAHPS analysis program using the SAS macro language.

  • P01: Integrated Care for Dual Eligibles

    Project Lead: David Grabowski
    Medicaid

    Background:

    Medicare beneficiaries also eligible for Medicaid -- the "duals" -- are a heterogeneous, vulnerable, high-cost, high-priority group. Duals are on average sicker than Medicare beneficiaries not eligible for Medicaid, and often have multiple chronic conditions. Conflicting regulations and incentives between Medicare and Medicaid fragment care, reduce quality, and increase total spending. The dually eligible with their complex needs are especially likely to benefit from coordinated care, yet this group is less likely than other Medicare beneficiaries to enroll in coordinated care plans. The juxtaposition of high need and cost, an inefficient Medicare/Medicaid partnership, and over-reliance on fragmented care models creates an opportunity for policy to both improve care and economize on public funds.

    This project focuses on the Medicare Advantage (MA) Special Needs Plans (SNPs). SNPs were authorized under the Medicare Modernization Act (MMA) of 2003 with the idea of attracting a different type of beneficiary into MA. Enrollment has grown steadily in SNPs and today roughly 1.8 million individuals are in these programs. We will exploit national payment changes as a series of natural experiments to explain the impact of payment policies on enrollment and care outcomes in MA SNPs for dually eligible beneficiaries.  Moreover, we plan to study how differences in Medicaid payment policies across the states affect enrollment and care outcomes in MA SNPs for dually eligible beneficiaries. Moreover, we will conduct a series of qualitative interviews with MA SNPs in order to provide further insight into the delivery and care models to understand access to care and the impact on care outcomes.

    This project is a part of the P01 grant that focuses on the successful integration of financing and care for Medicare.

  • P01: Effects of ACOs in Medicare on Utilization and Quality

    Project Lead: Michael McWilliams

    Background:

    In principle, delegating risk should encourage large integrated provider groups to achieve efficiencies. As suggested by our prior work, risk-bearing plans in MA may produce greater value than TM in terms of quality of care and resource use, but because MA plans are typically not clinically integrated with providers, the influence they can exert on contracting providers may be limited relative to ACOs. In Project 3, we will conduct rigorous evaluations of ACO initiatives in TM to determine if direct partial risk contracting with integrated provider groups is a viable complementary strategy for Medicare to control spending while improving quality of care. The gains achieved by an ACO will be determined by the changes in payment incentives introduced through its contracts with Medicare and other payers and its capacity to limit utilization and improve quality of care in response to those incentives. As suggested by previous research, the ability of ACOs to deliver more cost-effective care may be related to its structural characteristics such as size, specialty mix, and integration with hospitals. In particular, advanced models of primary care such as the patient-centered medical home have been proposed as essential building blocks of high-performing ACOs. As potential predictors of performance under new payment incentives, these factors may also influence organizations’ decisions to participate in the Medicare ACO programs.

    The project will identify conditions systematically related to effective responses by organizations to ACO payment models. In this project, we focus on organizations participating in the Medicare ACO programs because of their large number, their diversity, and the concentration of concurrent commercial ACO contracts among them. We will link novel national databases on provider organizations, their structural capabilities, and their commercial ACO contracting to claims data to identify and describe ACOs and non-ACO provider groups. By elucidating predictors of program participation and responses by organizations, our project will provide an empirical basis for fostering organizational learning from high performers, improving the structure of ACO contracts, and estimating potential gains from program expansion to existing and newly integrated provider groups. By assessing spillover effects of risk contracts in Massachusetts on ACOs’ patients not included in those contracts, our project will also characterize the extent of organizational change elicited by mixed payment incentives and the potential benefits of aligning incentives across payers.

    This project is a part of the P01 grant program that focuses on the successful integration of financing and care for Medicare.

  • P01: Utilization, Quality, Selection and Prices in Medicare Advantage

    Project Leads: Joseph Newhouse and Bruce Landon

    Background:

    Even with the advent of ACOs, for the foreseeable future the MA program is likely to remain numerically the most important alternative to Traditional Medicare (TM). Although MA currently enrolls nearly a third of Medicare beneficiaries, potential reimbursement reductions could diminish the array of supplemental benefits offered to MA beneficiaries, increase premium and cost sharing levels, and decrease the willingness of both beneficiaries and plans to participate. Moreover, many of these changes could also affect selection, the management of care, and the prices private MA plans negotiate with health care providers with repercussions for the entire delivery system. Given the prominence of MA in the Medicare program, evaluating these critical aspects of MA in light of these policy changes represents an important opportunity for research. 

    This project continues our study of the provision of services, selection, and the quality of care under MA and how all of those change over time as new policies are implemented. By continuing the complementary stock and flow comparisons of health risks in the MA and TM populations that we have employed using both administrative and survey data, as well as margin data from MA plans, we will clarify how payment changes affect risk selection and whether selection patterns become more socially efficient. Through the incorporation of data from Truven MarketScan as well as from several additional health plans that serve this market, we will be able to examine for the first time pricing MA plans obtain from providers and how those prices change over time and relate to market-level factors such as provider concentration. We also will extend our work comparing the provision of services and quality of care both over time and to additional clinical areas, including comparisons of drug treatment patterns for chronic diseases. In short, in our current project we have developed methods to address the salient policy issues that MA poses, we have addressed them at a time when reimbursement was becoming more generous and MA was expanding, and now we can apply our methods to a more austere reimbursement policy in which some successful ACO plans may nonetheless decide to convert to MA plans. 

    This project is a part of the P01 grant program that focuses on the successful integration of financing and care for Medicare.

Mediciad

  • Effect of Medicaid Managed Care on Spending Levels and Predictability

    Project Lead: Timothy Layton

    Background:

    The primary goal of this project is to estimate the effect of a state switching from fee-for-service Medicaid to Medicaid Managed Care on the level of Medicaid spending and the predictability of Medicaid spending at the state level. This data is de-identified and will eventually be available for public.

  • Determinants of Medicaid FFS and MMC Performance in New York State

    Project Lead: Timothy Layton

    Background:

    This research will assist the New York State Department of Health (NYSDOH) in efforts to improve the operation of managed care organizations (MCOs). Our work has several aims. First, we will conduct a rigorous evaluation of strategic MCO behavior from 2007-2012. Second, we will examine the impact of a Medicaid Physician Fee Schedule increase in 2009. Third, we will examine the impact of the phased implementation of Enhanced Ambulatory Patient Groups (EAPGs) for outpatient services. Each of these aims is thematically-linked by an effort to improve the administration of MCOs by providing vital information on plan and provider behavior to state policymakers. The analyses will be shared with the NYSDOH and submitted to peer-reviewed academic articles for publication.

    NOTE: all data users must complete HIPAA data training. CITI provides elective module on HIPAA. Certification must show module.

  • P01: Integrated Care for Dual Eligibles

    Project Lead: David Grabowski
    Medicare

    Background:

    Medicare beneficiaries also eligible for Medicaid -- the "duals" -- are a heterogeneous, vulnerable, high-cost, high-priority group. Duals are on average sicker than Medicare beneficiaries not eligible for Medicaid, and often have multiple chronic conditions. Conflicting regulations and incentives between Medicare and Medicaid fragment care, reduce quality, and increase total spending. The dually eligible with their complex needs are especially likely to benefit from coordinated care, yet this group is less likely than other Medicare beneficiaries to enroll in coordinated care plans. The juxtaposition of high need and cost, an inefficient Medicare/Medicaid partnership, and over-reliance on fragmented care models creates an opportunity for policy to both improve care and economize on public funds.

    This project focuses on the Medicare Advantage (MA) Special Needs Plans (SNPs). SNPs were authorized under the Medicare Modernization Act (MMA) of 2003 with the idea of attracting a different type of beneficiary into MA. Enrollment has grown steadily in SNPs and today roughly 1.8 million individuals are in these programs. We will exploit national payment changes as a series of natural experiments to explain the impact of payment policies on enrollment and care outcomes in MA SNPs for dually eligible beneficiaries.  Moreover, we plan to study how differences in Medicaid payment policies across the states affect enrollment and care outcomes in MA SNPs for dually eligible beneficiaries. Moreover, we will conduct a series of qualitative interviews with MA SNPs in order to provide further insight into the delivery and care models to understand access to care and the impact on care outcomes.

    This project is a part of the P01 grant that focuses on the successful integration of financing and care for Medicare.

Skilled Nursing

  • Intervention to Manage Acute Changes in Home Care Patients

    Project Lead: David Grabowski

    Background:

    Through a unique partnership with ClearCare, a home care software company, and Right At Home, a national home care company with over 14,000 clients, we propose a randomized evaluation of an Intervention in Home Care to Improve Health Outcomes (In-Home). In-Home is a low-tech, low-cost telephonic-based checklist administered by home care caregivers to patients, which assesses for acute changes in a patient’s physical or cognitive clinical status. The In-Home checklist tool will allow changes in clinical status to be relayed in real-time to a clinical case manager who will review the home care recipient’s condition and help decide the appropriate course of action (e.g., schedule a visit with the individual’s primary care physician). By identifying avoidable conditions in their early stages, we hypothesize that In-Home will prevent avoidable hospital admissions and improve overall health outcomes. Because it is low-tech and low-cost, we hypothesize that the tool will be easily used by caregivers and that savings from prevented hospitalizations will more than offset the costs of the program.

    Specific Aims:

    1. Access how the In-Home Program impacts all-cause mortality, and mortality from avoidable causes
    2. Access how the In-Home Program impacts the rate of all-cause hospitalizations and avoidable hospitalizations
    3. Assess the impact of the In-Home Program on spending and health, computing the return on investment associated with use of the tool.
  • Impact of Enhanced Primary Care in Nursing Homes

    Project Lead: David Grabowski

    Background:

    Many long-stay nursing home residents have very poor access to primary care, which often leads to unnecessary health care utilization and poor health outcomes. The Evercare Model, offered by UnitedHealthCare as a Medicare Advantage (MA) plan to nursing home residents, provides a treat-in-place model of care for enrollees through the use of nurse practitioners (NPs). The objective of this study is to understand the impact of the Evercare Model on outcomes for long-stay nursing home residents, including emergency department and acute care inpatient utilization, rates of readmission, as well as the amount Medicare spends on care. Through a unique partnership with UnitedHealthCare, we will have data on Medicare utilization and spending for Evercare Model enrollees. Using the 5% sample of Medicare beneficiaries in fee-for-service (FFS) Medicare, we will identify a similar set of individuals in the local market that is equivalent across patient demographics and clinical characteristics of the Evercare group. After matching, we will use multivariate regression models to compare Medicare spending and utilization across the Evercare plan and FFS Medicare beneficiaries. This study will provide the first large-scale evaluation of whether an MA plan with an increased clinical presence can improve outcomes, which may have a profound impact on the delivery of services to the nearly one-million long-stay nursing home residents in this country. These results, which will be of interest to CMS, MA plans, nursing homes, and Medicare beneficiaries, will be shared widely through publication in peer-reviewed journals, presentations to stakeholders, Capitol Hill briefings, and other outlets.

  • Specialization in Nursing Home Care

    Project Lead: David Grabowski

    Background:

    This project has three aims:

    1. To quantify the level, variation and correlates of quality for persons with mental illness or dementia in nursing homes. We describe the variation in quality of general health care (such as in functioning improvement) and mental health care (such as rates of antidepressant use) and test a series of hypotheses about the variation. We partition the variation into within and across facility components.
    2. To quantify the impact of nursing home specialization on quality of mental health care for residents with a mental illness or dementia.
    3. To analyze the impact of nursing home specialization on quality of care for residents without a mental illness or dementia.
  • Role of ownership in the provision of hospice care

    Project Lead: David Grabowski

    Background:

    The hospice provider market has changed markedly over the past 30 years, transitioning from a relatively small base of locally-run, non-profit agencies to a larger market where a slight majority of agencies, some with a national presence, are run on a proprietary basis. In the context of Medicare hospice spending more than quadrupling over the last decade, policymakers have paid particular attention to the emergence of a robust for-profit hospice sector and increased use by nursing home residents, raising questions about the extent to which some agencies aggressively target more profitable patients.1 To date, however, there have been limited studies looking at the role of for-profit vs. not-for-profit hospice agencies (FP vs. NFP), and all of these have focused on these two ownership categories in the aggregate. The general tenor of these findings has been that FPs disproportionately enroll patients with non-cancer diagnoses, nursing home residents, and longer length-of-stay hospice users.

Technology

  • Technology Diffusion under New Payment and Delivery Models

    Project Lead: Sharon-Lise Normand
    Medicare

    Background:

    This study intends to examine the relationship between organizational traits and diffusion of new medical technology, and to explore the impact of new health care delivery models on costs and quality of health care associated with the new technologies. Study objectives include the comparison of the speeds and mean adoptions times across types of technologies (drugs, devices, and biologics) within disease areas, across lower and higher value technologies, within and across organizations. Our overall objective involves providing quantitative summaries of the associations of new risk-based payment models with fundamental decisions about technology adoption at the organizational level.

    In Aim 1 we will study the diffusion (speed, mean adoption times, ceilings) of selected new technologies as a function of organizational characteristics to determine whether decisions to use new technologies are correlated among different technology types, or within and between disease conditions, within and between organizations.

    In Aim 2, we will distinguish higher from lower value services, identify organizational factors predictive of their use, and determine if organizational decisions to adopt higher vs. lower value services are correlated. Using the findings from Aims 1 and 2,

    Aim 3 will assess the impact of the Medicare risk-based sharing arrangements on total spending, new technology spending, and spending on higher and lower value services, adjusting for organizational factors including organizational-specific diffusion propensities.

  • Technology Diffusion Pathways

    Project Lead: Sharon-Lise Normand

    Background:

    There are a number of new strategies with the goals of containing spending growth and to improve the quality of care to patients.  The implementation of these strategies encourage provider organizations to meet the spending and quality of care se goals through delivery system changes.  Adoption of new medical technology is a primary driver of health care spending growth.  Physicians and the organizations in which they practice have various options to adopt and use new and existing technologies.  This study conducts an in-depth investigation of the patterns of diffusion of new technologies, including drugs, biologics, and devices, across organizations of different types, and assess the effects of a new model for organizing and financing health care. We will compare rates of adoption and use, across types of technologies within disease areas, across lower and higher value technologies, and across organizational forms.

Other Topics

  • Evaluation of Value-Based Payment Reform in Arkansas

    Project Lead: Michael Chernew
    Cost Policy

    Background:

    Much of the innovation in payment is occurring at the state level. Some of these approaches focus only on Medicaid, others on multi-payers. Some rely on global payments and others on bundled payments for selected episodes. Evidence about the impact of state level reforms is lacking. We will evaluate the Arkansas Health Care Payment Improvement Initiative (AHCPII), an attempt to move the state’s public and private payment systems toward value-based purchasing. AHCPII consists of two major components: an upside-only population-based payment model, which utilizes a patient-centered medical home (PCMH) approach, and a retrospective episode-based payment model. In short, we will evaluate how a unique PCMH-based shared savings program – which is an attempt at both Medicaid and private payment reform – is impacting the cost and quality of care in a state with fragmented primary care delivery. We will use the Arkansas All-Payer Claims Database (APCD) - Medicaid claims data for years 2013-2016 which include detailed information on medical care utilization and spending. We hope to inform future attempts at payment transformation in states with environments similar to Arkansas (e.g. multi-payer involvement and fragmented providers). Importantly, we are also building an analytical framework that can be readily applied to perform rigorous, systematic evaluations of other state-based payment reforms. Arkansas is a leading example that statewide, primary care-based payment reform can be implemented. We therefore believe that our research will contribute to the broader debate over payment reform models.

  • Adverse Selection in Regulated Health Insurance Markets

    Project Lead: Tim Layton
    Selection

    Background:

    Private health insurance markets are plagued with problems stemming from adverse selection, or the tendency of sicker consumers to choose plans offering more generous coverage. While the theory behind adverse selection problems is well-established, their practical importance is less well-known, especially in individual insurance markets such as the new Marketplaces created by the Affordable Care Act (Connect for Health Colorado in Colorado), the Medicare Advantage program, and Medicaid Managed Care. These markets all include mechanisms to combat adverse selection problems, the most prominent of which is risk adjustment, a plan payment mechanism that adjusts the payments plans receive according to the health status of their enrollees.
    In this project, we aim to study the extent of adverse selection problems in these three markets, the Colorado ACA Marketplace, Medicare Advantage, and Medicaid Managed Care. We will first assess how severe selection on health status appears in these markets. We will then assess how well risk adjustment transfers/subsidies compensate plans for differences in health status and explore how alternative payment policies might improve the functioning of these markets. We will also explore the sources of the patterns of selection, focusing specifically on provider networks, drug coverage, and measures of plan quality. Finally, we will study interactions of adverse selection with insurer and provider market power.