CMS’s role in the U.S. health care mosaic will be pivotal as the nation shifts to improved approaches for organization, payment, consumer engagement, understanding of the bioscience foundations of health, and data management for the provision of quality, equitable health care. This transition will take place over many years, but some key shifts are already underway, with some dimensions being planned while others evolve on their own. At every stage, the capacity to improve decision making throughout the entire system will depend not only on having timely access to data but also on the capacity to transform the raw data first into information and ultimately into intelligence to support future planning and action.
Data warehouses, business intelligence, and data analysis have existed for more than 30 years and have a long history of use in the sciences. The explosive growth of data in all areas of business, science, medicine, and life in general has opened ever greater potential for discovery and understanding through analysis of data. Over the past 20 years the advent of virtual or data-driven science has meant that in some areas it is possible to experiment or discover for the cost of database searches and analysis. Data-driven techniques can have application in medicine and health care in addition to sciences such as astrophysics. For example, if data from longitudinal studies, clinical observations, and other health care activities were made available to researchers, studies of the comparative effectiveness of alternative medicines could be conducted in a fraction of the time and cost required for clinical trials, which are often extremely
limited in terms of the population studied, are expensive, and can incorporate biases.1
Another health care trend that has implications for CMS’s data-related efforts is the ongoing shift of individual practitioners from solo and/or small group practices into care systems and networks. This realignment of providers and institutions could result in far more payments for bundled services intended to achieve defined outcomes. The efficiency and effectiveness of such an approach will depend on the collection and transfer of a great deal of data.
CMS is in the process of transforming itself to enable a focus on information collection and data management while still fulfilling its traditional mandates, including retrospective payment for health services for segments of the population. Several trends in health care illustrate the broad need for a more data-centric approach, including the diffusion of electronic health records (EHRs), practitioner positioning into care networks, and increased consumer access to and demand for health and medical information.
In aggregate, these trends in health care regarding data will interact in ways to produce both additional work and new requirements for CMS. Although the ultimate result is currently unknowable, the drive to achieve great value for health care for both individuals and populations is not likely to abate anytime soon, especially in light of the demographic pressures and the size of the financial investment the nation is making in health care services.
Data are essential to and underpin nearly all of the efforts CMS is undertaking—and data are an essential driver for the strategic technology plan advocated in Chapter 2, motivate the meta-methodology outlined in Chapter 3, and are a key impetus for the organizational changes discussed in Chapter 4. Gathering these data and sorting out how to make data available and to whom cannot be envisaged adequately until all stakeholders have been engaged and are contributing to an ongoing discussion. Doing that incremental engagement is part of the committee’s recommended approach and is essential to the development of mechanisms for future data management as an aspect of CMS’s new and changing relationship to data and information. In addition, CMS has to grapple with ensuring that only authorized users have access to data such as personal health information or other individual-level information.2
1 Sharon Begley, 2011, “The Best Medicine: A Quiet Revolution in Comparative Effectiveness Research Just Might Save Us from Soaring Medical Costs,” Scientific American 305(July):50-55.
2 While posing technical challenges, the question of who is authorized to access what sorts of data is a policy matter distinct from the technical challenges.
This chapter discusses several ways in which the data and information collected by CMS are used extensively within the agency for purposes such as quality-related efforts, policy analysis, and combating fraud, as well as for informing consumers and managing payments. It also addresses recent legislative mandates for CMS—such as use of CMS information to analyze racial and ethnic disparities so as to contribute to their reduction.
Heightened since the publication of a 2000 Institute of Medicine report,3 the effort to improve the quality and safety of health care has been fostered by CMS and many other key actors in health care, including insurers, professional associations, accreditation and review groups such as the Joint Commission,4 the National Committee for Quality Assurance (NCQA),5 and the National Quality Forum (NQF),6 and care providers such as physicians and hospitals. Although this effort has been uneven and at a slower pace than hoped for by safety champions,7 the cumulative effect of several factors—policy maker and academic attention to quality and safety improvement,8 Medicare payment and reporting incentives such as pay for performance9 and value-based purchasing, and the proliferation of EHRs, aided by the financial incentives in HITECH—have
3 Institute of Medicine, 2000, To Err Is Human: Building a Safer Health System, Washington, D.C.: National Academy Press.
4 Joint Commission Center for Transforming Healthcare, 2011, “About the Center,” website, available at http://www.centerfortransforminghealthcare.org/about/about.aspx, last accessed August 8, 2011.
6 National Quality Forum, 2011, “ABC’s of Measurement,” website, available at http://www.qualityforum.org/Measuring_Performance/ABCs_of_Measurement.aspx, last accessed August 8, 2011.
7 Mark R. Chassin, Robert W. Galvin, and the National Roundtable on Healthcare Quality, 1998, “The Urgent Need to Improve Health Care Quality,” Journal of the American Medical Association 280(11):1000-1005.
9 CMS, 2011, “IPPS Regulations and Notices,” website, available at http://www.cms.gov/acuteinpatientpps/ipps/itemdetail.asp?filterType=none&filterByDID=-99&sortByDID=4&sortOrder=ascending&itemID=CMS1229138&intNumPerPage=10, last accessed August 8, 2011.
created an environment in which the measurement and improvement of health care10 are better understood now than 10 years ago.
The budgetary imperatives to “bend the cost curve”11 and the movement for improved quality come together in the Patient Protection and Affordable Care Act (PPACA), which contains numerous initiatives (for example, the promotion of accountable care organizations and the creation of the CMS Innovation Center) intended to result in better, more cost-effective care for Medicare beneficiaries. CMS has substantial responsibility in implementing parts of the PPACA and therefore in creating the standards, reporting mechanisms, payment processes, and data and measurement requirements that will foster these hoped-for improvements in health care for beneficiaries and more broadly for the U.S. population.
Nationwide investment in EHRs and EHR systems—enabled through HITECH—is occurring on such a broad scale that it has the potential to accomplish some sea changes in the information infrastructure of health care delivery across the country and to generate the fine-grain data needed for improved health care delivery. For example, although the focus today is heavily on clinical transactions and data capture, insurance recipients will increasingly have the ability to access information about their health directly through secure patient portals. The day is coming when evidence-based care protocols will support not only decision making by clinicians and patients but also direct enhancements of health care quality and safety for diverse populations and subpopulations through IT-based applications that help ensure quality as care is being delivered rather than focusing on measurement after the fact. The HITECH approach to meaningful use of health information technology also has the potential to allow measurement of the quality of performance without engaging data abstractors, thereby making such assessment much less expensive. However, there are also some limitations that will have to be overcome— including those related to the connectivity of interoperable systems, to systems themselves, to users, and to potential barriers resulting from public policy and regulation. The shift will have clear implications for CMS.
The data provided by the expansion of EHRs deserves special notice. Although policy analysts (in, for example, the Medicare Payment Advisory Commission (MedPAC), the Congressional Budget Office, the Office for Management and Budget, and research-oriented “think tanks”) have used Medicare claims data for decades in proposals for improving the
10 Atul Gawande, 2007, Better: A Surgeon’s Notes on Performance, New York: Metropolitan Books.
11“Obama Talks Health Care with Fred Hiatt,” 2009, Washington Post, July 22, available at http://www.washingtonpost.com/wp-dyn/content/article/2009/07/22/AR2009072202522.html, last accessed August 1, 2011.
health care system, the more robust data contained in EHRs now allow for richer analyses of the health care provided to populations than those analyses based on data previously available. Because the results of medical procedures, medications, and treatments can be measured and analyzed by using the information contained in EHRs, providers such as integrated medical groups and major hospital systems, which have led the way in implementing EHRs, are now better able to improve care processes by, for example, introducing evidence-based “alerts” and guidance for physicians and nurses during the actual provision of medical care. In addition, having electronic access to real-time health and medical data can advance people’s capacity to manage health care.
CMS’s role in the context of today’s data explosion is multifaceted. CMS has responsibility for establishing and evaluating the meaningfuluse standards and incentive payments legislated in HITECH. It establishes standards for quality reporting, such as the “core measures” required in the value-based purchasing mandated by the PPACA. Through accreditation processes, CMS can measure how well providers meet the “conditions of participation” in Medicare. It can set the criteria by which quality improvement efforts (such as “medical homes”) are evaluated by patient-specific clinical data. The possible ways that such data in electronic records can be used to improve measurement and payment processes are numerous, and CMS will have to determine the preferred options within its broader implementation of PPACA.
This transition will induce heavy demands for the capture of accurate, meaningful data that account fully for the health status of those served by health care delivery organizations. To manage this accounting, CMS is likely to focus increasingly on the full range of social determinants of health status, moving beyond those that relate solely to health care technology and medical interventions. An example of the complexity of this effort can be seen with respect to the use of billing data for quality measurement; although some relevant information can be extracted, that data alone is not enough. Future efforts are likely to require ongoing attention to high-resolution information in the form of natural language or formalized data flows realized through an evolution of ontologies, terminologies, and, ultimately, relevant standards that can help to ensure that meaning is not lost in the translation of data to understanding.
Clinical outcomes data are currently used throughout the health care system to monitor, improve, and report on the quality of care in a wide variety of settings. Increasing use of EHRs, and the potential associated increase in available clinical data, offer both great potential benefits in terms of measuring and monitoring quality—and potential risks in terms of cost, acceptability, and protection of patient privacy. CMS will be tasked not only with using outcomes data to evaluate the care of its own benefi-
ciaries but also with building links to information systems with data on the care given to others so that comprehensive evaluations of the quality of care provided for individuals and groups can be developed, both inside and outside government programs.
There are a number of measures of basic quality for which data could be collected on all providers. Additional information from certain geographic regions or practice settings might then be collected, if problems arise, to help illuminate the source of a problem. The committee is aware of the assumption of some that CMS should plan to collect information on patients following what Diamond and colleagues call the “dominant paradigm” for handling population health data: “gather copies of all the detailed information one needs, normalize the information once one has it, and then run queries against that data storehouse.”12 Such an approach—if applied to CMS’s role in analyzing and monitoring health care quality, equity, and safety—has drawbacks in terms of cost and the potential for violations of privacy, and may reduce the acceptability of EHRs for many practitioners. In addition, such an approach is relatively rigid, requiring advance knowledge of what data are needed and implying a single national approach to improvement of quality.
Others argue for a distributed analytic system that, once in place, could be used to increase monitoring in settings where problems have been identified while maintaining only minimum information on practice groups that are functioning well. A distributed system could be deployed rapidly in support of local efforts at containment of public health emergencies such as the Severe Acute Respiratory Syndrome epidemic. Such an approach, in which information remains at the source where it was collected, is being used increasingly for purposes as disparate as public health surveillance and cancer research. Of particular interest for CMS’s purposes is the Distributed Research Network (DRN) supported by the Agency for Health Care Research and Quality (AHRQ), which is designed to support composite data analysis on the safety and effectiveness of health care.13 Distributed approaches have disadvantages as well, such as decreased access to data for some stakeholders and potentially less comprehensive analytic capabilities.
Separate from how data are collected and stored are the many opportunities clearly afforded in the area of quality and safety by the development of effective analytics. Information from payers other than Medicare
12 Carol C. Diamond, Farzad Mostashari, and Clay Shirky, 2009, “Collecting and Sharing Data for Population Health: A New Paradigm,” Health Affairs 28(2):454-466.
13 Andrew J. McMurry, Clint A. Gilbert, Ben Y. Reis, Henry C. Chueh, Isaac S. Kohane, and Kenneth D. Mandl, 2007, “A Self Scaling, Distributed Architecture for Public Health, Research and Clinical Care,” Journal of the American Medical Informatics Association 14(4):527-533.
and Medicaid could be analyzed in combination with CMS information to provide a much more comprehensive view of the performance of a practitioner, group, or system. Information from sources other than individual clinical records (such as registries with data on the incidence of exposure to disease) can, when appropriate, be included in the analysis to give a clearer picture of trends in uses of medical care. In planning its future quality management strategies, CMS will have to resolve for itself and in collaboration with its stakeholders what strategies for handling such data it will adopt.
Outside researchers, many of whom are investigating quality-related research questions, currently make extensive use of the data sets generated by CMS. Although in comments received by the committee the cost of obtaining CMS data was raised as a concern, the chief complaint was that currently almost 2 years elapse before the data can be accessed. For example, the most current data available on the frequently used MEDpar file (hospital discharges) is from 2009, with the release of data from 2010 expected in October 2011 as of this writing. Data in the “access to care” section of the Medicare Current Beneficiary Survey, another area of particular interest to many, is also available only through 2009, with the 2010 update scheduled for the summer of 2012.14
Among the reasons for these delays are that claims data from any insurance program are not complete until sometime after the date of service. In Medicare’s case this “claims lag” is fairly short, with 98 percent of claims submitted within 3 months,15 although even so, a data set for any given year still lacks some claims at the end of the first quarter of the following year. To permit earlier release of claims, CMS could use a variety of strategies, such as providing an interim data set of the mostused information on a quarterly basis with the limitations clearly spelled out. Many of the modernization steps discussed throughout this report will make data integration easier (for example, integrating the reports from Medicare Managed Care with those from fee for service), leading toward earlier release. Much earlier release of survey data, which should be technically possible even now, will support the best use of this important information.
14See http://www.resdac.org/Tools/TBs/TN_015_CMS%20Data%20Availability_508%20.pdf, last accessed June 14, 2011.
15 Department of Health and Human Services, CMS, 2011, “Medicare Program; Medicare Shared Saving Program: Accountable Care Organizations,” Proposed Rule, Federal Register 76(67; April 7):19554.
CMS’s first effort at consumer-oriented, public reports on the quality of care first began approximately 15 years ago with dialysis units. Those results are available at “Dialysis Compare”16 (with individual data sets available on data.gov), and the effort is widely regarded as successful. It is not clear that consumers have used the information extensively—but providers pay attention and work to meet the standards. Dialysis, however, has two characteristics that make it unique: (1) Medicare is, for all practical purposes, the only payer, and so data on Medicare beneficiaries reflect the full experience of dialysis centers, and (2) dialysis has a limited number of easily measurable objective outcomes.
CMS’s efforts have expanded to other consumer-oriented “report cards,” and the CMS website now also has sections, known as “Hospital Compare”17 and “Nursing Home Compare,”18 that make use of information collected in surveys as well as reporting of quality measures and “never events” (adverse outcomes that ought not to have happened, such as wrong-site surgery). There are limitations to the information’s utility, because hospitals and nursing homes serve many non-Medicare beneficiaries, and so even the most precise analysis of care received under Medicare may not reflect overall performance. However, as CMS’s ability to use more granular data from sources outside the organization becomes more robust, these reports have the potential to become more accurate and, consequently, more useful.
Although the mechanisms for consumer engagement are somewhat unclear, more groups and individuals will seek greater access to CMS data and information. It is reasonable to assume that the equivalent of citizen engagement in health services and health policy research will increase, analogous to social networking in other domains, as will online dialogue relating to the output of such efforts.19
One of the most important secondary uses of CMS data on health care encounters is the analysis of current spending patterns and projections of future spending. The number of reports that make use of CMS encounter data is vast; two of the most important are the annual trustees’ report
19 Schumpeter, 2011, “Saving Britain’s Health Service: The NHS Needs to Learn from Innovations in the Rest of the World,” The Economist, June 16.
(which evaluates the current status of the trust funds20) and the “data book” (published at least annually by the MedPAC21). The trustees’ report focuses on projections of future costs of the Medicare program, and the data book is a more detailed analysis of changes in patterns of use and spending over time.
These documents use information from a variety of sources; the most significant of these are (1) the Medicare Current Beneficiary Survey, which is a continuous, multipurpose survey of a nationally representative sample of aged, disabled, and institutionalized Medicare beneficiaries, and (2) the various “market baskets” developed by an economic forecasting firm22 to serve as the basis for the annual updates of payments to hospitals and other providers.
CMS and MedPAC are not the only government and quasi-government agencies using encounter data—other groups, such as the Congressional Budget Office, the Government Accountability Office, and HHS’s Office of the Inspector General, also depend on encounter data for their analyses and predictions. The Independent Payment Advisory Board, established by the PPACA, will also require Medicare data in order to fulfill its mission to help reduce the rate of growth in Medicare costs without affecting coverage or quality.
As the Medicare actuaries note in their discussion of the data in the trustees’ report, there are elements in the information, such as delayed decisions on the exact amounts paid to specific hospitals, which lead to small error rates, which are multiplied when extended projections are developed.23 It is therefore particularly important that the same sources of information be available to independent researchers in order to facilitate well-informed debate regarding the future of the Medicare and Medicaid programs.
20 The Boards of Trustees of the Federal Hospital Insurance and Federal Supplementary Medical Insurance Trust Funds, 2011, 2011 Annual Report of the Boards of Trustees of the Federal Hospital Insurance and Federal Supplementary Medical Insurance Trust Funds, available at https://www.cms.gov/ReportsTrustFunds/downloads/tr2011.pdf, last accessed July 21, 2011.
21 Medicare Payment Advisory Commission (MedPac), 2010, A Data Book: Health Care Spending and the Medicare Program, June, available at http://www.medpac.gov/documents/Jun10DataBookEntireReport.pdf, last accessed July 21, 2011.
22 Currently Global Insights, headquartered in Lexington, Massachusetts.
23 Boards of Trustees, 2011, “Actuarial Methodology,” 2011 Annual Report, p. 150.
Evidence confirms the reality of health disparities experienced by minority Medicare, Medicaid, and other beneficiaries served by CMS.24 These individuals constitute an ever-increasing percentage of the total— for example, nearly 20 percent of Medicare beneficiaries and 60 percent of all individuals receiving assistance through Medicaid.25 Persistent health inequities26 among population groups in the United States are not only unacceptable as characterized by the Institute of Medicine in its landmark 2003 report,27 but also costly,28 contributing substantially to the nation’s spiraling health care costs.
CMS’s key role in the transformation of the nation’s health care system has been noted throughout this report. The significance of that role in addressing health disparities is also critical. It will not be possible for CMS, and the nation as a whole, to cross the “quality chasm”29 and achieve the transformation of the nation’s health system if the needs of all populations are not addressed in an equitable manner.
The committee is aware that strategies to reduce health disparities are receiving high-priority attention by CMS, HHS, and the U.S. Congress.30
24 See, for example, David C. Goodman, Dhannon Brownlee, Chaing-Hua Chang, and Elliott S. Fischer, 2010, “Regional and Racial Variation in Primary Care and the Quality of Care Among Medicare Beneficiaries,” from the Dartmouth Atlas Project, available at http://www.dartmouthatlas.org/downloads/reports/Primary_care_report_090910.pdf, last accessed August 1, 2011; and Tracy Onega, Eric J. Duell, Xun Shi, et al., 2010, “Race Versus Place of Service in Mortality Among Medicare Beneficiaries with Cancer,” Cancer 116(11):2698-2706.
25 Kaiser Family Foundation, “Distribution of Medicare Enrollees by Race/Ethnicity, States (2008-2009), U.S. (2009),” available at http://www.statehealthfacts.org/comparebar.jsp?ind=297&cat=6, accessed August 1, 2011; and Kaiser Family Foundation, “Distribution of the Nonelderly with Medicaid by Race/Ethnicity, States (2008-2009), U.S. (2009),” available at http://www.statehealthfacts.org/comparebar.jsp?ind=158&cat=3&sub=42, last accessed August 1, 2011.
26 Agency for Healthcare Research and Quality, 2010, “Disparities in Healthcare Quality Among Racial and Ethnic Minority Groups: Selected Findings from the 2010 National Healthcare Quality and Disparities Reports,” available at http://www.ahrq.gov/qual/nhqrdr10/nhqrdrminority10.pdf, last accessed August 1, 2011.
27 Institute of Medicine, 2003, Unequal Treatment: Confronting Racial and Ethnic Disparities in Health Care, Washington, D.C.: The National Academies Press.
28 Thomas A. LaVeist, Darrell J. Gaskin, and Patrick Richard, 2009, The Economic Burden of Health Inequalities in tfie United States, Report by the Joint Center for Political and Economic Studies, available at http://www.jointcenter.org/sites/default/files/upload/research/files/The%20Economic%20Burden%20of%20Health%20Inequalities%20in%20the%20United%20States.pdf, last accessed August 1, 2011.
29 Institute of Medicine, 2001, Crossing the Quality Chasm: A New Health System for the 21st Century, Washington, D.C.: National Academy Press.
30 U.S. Congress, 2009, “Addressing Health Care Disparities,” Congressional Record, Sec. 1946, November 19, S11734.
It has also noted the findings of agencies and other observers, both within and outside government, that the availability of data, disaggregated by race, ethnicity, primary language, and other factors, is essential for the accomplishment of this goal. Available evidence indicates, however, that adequately categorized, complete, and comprehensive data, collected by systematic and effective means, currently are not readily available at CMS.
CMS race/ethnicity data are of uneven quality with respect to accuracy and completeness, as documented by reports produced by HHS, IOM, and other agencies, as well as testimony received by the committee. For example, although the Social Security Administration (SSA) modified its data collection practices in 2008 to follow the categorization standards promulgated by the Office of Management and Budget (OMB), the updated procedures apply only to new Social Security and Supplemental Security Income claims and replacement number and lost card applications. These revised OMB standards, which provide for detailed race/ethnicity categories, do not apply to applications filed before 2008 or to applications received under SSA’s Enumeration at Birth process, which precludes the collection of race and ethnicity data because of states’ restrictions.
Recent developments, however, are encouraging. Although CMS’s report to Congress as required under the Medicare Improvements for Patients and Providers Act (MIPPA) of 2008 was not available for the committee’s review, it is expected to reflect new and innovative “approaches… for identifying and collecting and evaluating data on health care disparities on the basis of race, ethnicity, and gender” as mandated by the MIPPA.31 CMS’s requirement of those receiving EHR meaningful-use incentives to collect data on race, ethnicity, primary language, and other factors is an opportunity to “connect the dots,” revealing and tracking health care patterns and trends by population and subpopulation in relation to the quality of services received—not just for CMS beneficiaries but also for a much wider population of health care consumers.
CMS leadership also gave presentations to the committee on the primacy of equity in advancing a health quality agenda.32 And indeed, the value of these data has been succinctly described by HHS Secretary Kathleen Sebelius, who stated in her March report to Congress: “Improvements in the way data is collected help to pinpoint and address where
31 Medicare Improvements for Patients and Providers Act of 2008, HR 6331, 110th Congress, 2nd session.
32 Terris King, CMS Office of Minority Health, 2011, “Health Disparities,” presentation to the committee, April 18, Baltimore, Md.
health disparities exist.”33 CMS’s role in collecting and reporting race/ ethnicity, gender, and other disaggregated health data is critical to the equitable delivery of quality health services to all CMS beneficiaries. Thus CMS’s vision, strategies, and priorities for the use of information technology as well as its organizational and strategic technology plans will have to take this role and its requirements into account.
CMS’s Center for Program Integrity faces a formidable challenge in dealing with fraud, waste, and abuse within the Medicare and Medicaid programs. The following major elements of criminal fraud were identified for the committee:
• Registration of fraudulent providers and/or suppliers, most commonly involving providers of durable medical equipment;
• Fraudulent use of an existing provider number;34
• Fraudulent, duplicative, or excessive billings by an existing provider who is also| delivering legitimate services;35
• Theft of beneficiary identification; and
• Fraud in which the beneficiary participates (for example, billings for expensive services not rendered, with profits split with the beneficiary).
At present, only a tiny minority of claims are reviewed prior to payment. But as has been noted by both the current administrator of CMS and the secretary of HHS,36 and as industry experience demonstrates,37 it is materially more productive and efficient to identify questionable billings in advance of payment. An ability to analyze all claims prior to payment as a basic element in CMS’s fraud detection would enable unusual patterns to be identified and holds to be placed on those that are most
33 Department of Health and Human Services, Office of the Secretary, Office of the Assistant Secretary of Health, Office of Minority Health, 2011, Report on Minority Health Activities as Required by the Patient Protection and Affordable Care Act, PL. 111-148, available at http://www.healthcare.gov/center/reports/minorities03252011a.pdf, last accessed August 1, 2011.
34 For example, use of a retired physician’s identity.
35 See, for example, Mark Schoofs and Maurice Tamman, 2010, “In Medicare’s Data Trove, Clues to Curing Cost Crisis,” Wall Street Journal, October 25, available at http://online.wsj.com/article/SB10001424052748704696304575538112856615900.html, last accessed August 1, 2011.
36 See Healthcare Fraud Prevention Summit video, December 16, 2010, available at http://www.stopmedicarefraud.gov/videos/fraudprevention_boston.html, last accessed August 1, 2011.
37 Bob Shiflet, 2011, “Fraud Detection and Prevention in Large Scale Systems,” presentation to the committee, February 17, Irvine, Calif.
suspicious. This identification necessarily depends on techniques of pattern recognition across multiple data sets, including data sets related to claims, providers, patients, and third parties and including government and private data sources. CMS’s current information infrastructure was not designed to provide the ability to mine data in a timely fashion, even within a single system, much less across all of them.
Moreover, as certain patterns of fraud or abuse become identifiable and bad actors modify their strategies, CMS will need to continuously modify its approach and responses. As it is notoriously complex to retrofit legacy systems to deliver the agility and flexibility to meet these challenges, new information management and analysis solutions will have to be designed to provide this agility.
To prevent the registration of false providers and suppliers, CMS will have to develop strategies to deal with the 18,000 Part A and B provider enrollment applications and 900 curable medical equipment supplier applications received each month.38 At present the steps taken to ensure that only genuine providers and suppliers are enrolled include surprise site visits as well as a focus on high fraud areas. External data such as data on location are available on new business concerns. Using a modernized claims payment system, however, it would also be possible to conduct intensely focused pattern analysis of claims submitted by newly enrolled providers to detect outliers with high billing rates that can in turn be subject to on-site inspection. Similarly, pattern analysis of claims submitted by all providers for beneficiaries who have reached retirement age is likely to be productive in terms of identifying unusual and suspicious changes in billing behavior.
The current separation of the Medicare and Medicaid programs allows duplicate billings by the same provider for the same service. Merging that information in ways that allow detection of this sort of duplication would be useful. Using insurance exchanges to correlate data across all plans can also yield information about patterns of fraudulent activity that should make fraud detection more rapid and efficient. The committee heard arguments that reduction of fraud in the Medicare and Medicaid systems might be more easily accomplished through a preventive stance rather than an emphasis on detection and enforcement after fraudulent claims have been submitted.39 For example, with the increasing adoption of electronic health records, there are improved opportunities for detecting fraud when a patient is being scheduled or seen, or as a fee-for-service
38 CMS, 2010, “Partner with CMS,” website, available at https://www.cms.gov/Partnerships/Downloads/72010NMEPFraudandAbuse508.pdf, last accessed August 8, 2011.
39 Donald W. Simborg, 2011, “CMS IT and Fraud,” presentation to the committee, February 17, Irvine, Calif.
bill is being generated.40 Similarly, some of the metadata in EHRs could serve to identify patterns that suggest improper billings, such as notes written before the official date of service.
Although it is CMS that faces the onus of dealing with fraudulent claims when they are submitted, and the cost to the nation is generally judged to be enormous, partnerships with other agencies, and particularly the Office of the National Coordinator, may be required to effect some of the needed innovation. Some have argued that EHR vendors should be required to address these issues in their products, which raises the possibility of anti-fraud capabilities being incorporated into future criteria for meaningful-use payments. The Small Business Jobs Act of 2010, passed in September 2010,41 directs CMS to use predictive modeling and other techniques to identify improper claims and prevent the payment of such claims.42 CMS began using a new fraud management platform in July 2011.43 At present, however, the primary focus on fraud and abuse at CMS continues to be in the post-billing payment arena, where CMS has greater control but still faces significant challenges in recognizing fraud before bills are paid, after which funds can be recovered only with great difficulty.
In addition to criminal fraud, The CMS Center for Program Integrity must also monitor a number of complex rules about physician behavior. The anti-kickback statute44 and the physician self-referral statute45 forbid activities that may appear innocent to a new provider, such as the offer of a “medical directorship” or other position that involves generous pay-
40 D.W. Simborg, 2008, “Healthcare Fraud: Whose Problem Is It Anyway?” Journal of the American Medical Informatics Association 15(3):278-280; D.W. Simborg, 2011, “There Is No Neutral Position on Fraud!” Journal of the American Medical Informatics Association 18(5):675-677.
41 Public Law 111-240, Small Business Jobs Act of 2010, 124 Stat. 2504, September 27, 2010.
42 As stated by CMS, “The Small Business Lending Act, which was signed into law on September 27, 2010, included an anti-fraud provision requiring that CMS implement new software with “predictive modeling,” a type of analytical technology that already has been adopted in the credit card industry to identify potentially fraudulent bills. The provision requires CMS to launch a competitive bidding process by January 2011 for predictive modeling software contractors and to begin implementing the technology by July in the ten states with the highest Medicare fraud rates. A key driver to the success of Program Integrity (PI) at CMS is data integration—across programs and across patient, provider, and plan domains.” See CMS, 2010, “Modernizing CMS Computer and Data Systems to Support Improvements in Care Delivery,” December 23, available at https://www.cms.gov/InfoTechGenInfo/Downloads/CMSSection10330Plan.pdf, last accessed October 21, 2011.
43 CMS, 2011, “New Technology to Help Fight Medicare Fraud,” press release, June 17, available at http://www.cms.gov/apps/media/press/release.asp?Counter=3983, last accessed September 12, 2011.
44 Within the Medicare and Medicaid Patient Protection Act of 1987, 42 U.S.C. §1320a-7b.
45 Section 1877 of the Social Security Act, enacted in 1989, also referred to as the Stark law.
ment for little work. There is every reason to believe that good pattern analysis will be productive here as well.
Chapter 4 discusses issues related to internal data governance in CMS, and previous sections of this chapter discuss the potential utility that comes with the enormous amount of data that CMS already has, as well as data that will be generated in the future and will involve specific governance issues. As CMS prepares for a data-centric future, a number of questions will merit careful consideration. This section describes some of them, but the list not intended to be exhaustive.
• What is the scope of the data that will be available? The scope can range from national-level summary data to data with granularity at the level of states, counties, cities, individual institutions, specific providers, or even individual patients.
• What is the nature of the data to be provided? The data could be billing codes or could include overview-level clinical summaries. There might even be such specificity as clinical details and short-term outcomes. Even more specific would be data on long-term outcomes and the follow-up regarding patient status in the months or years after care, or long-range data on lifetime cumulative health status.
• Who will have access to the data? Although CMS itself and other payers or their proxies (for example, insurance companies, state Medicaid agencies, and so on) are among those that are likely to have first-order access, there will also be interest on the part of the providers and others from whom the data on quality and cost-effectiveness data are collected. In addition, as discussed in Chapter 1, an even broader range of potential data users includes academic researchers, public interest groups, certification bodies, and disease-focused societies that have an interest in CMS data. CMS might even choose to post some data sets (with suitable privacy protections) on data.gov and make them accessible to anyone.
• Different access and use models will have very different governance models. Access to the data of CMS and other payers would require relatively simple agreements and access authorization and authentication. Opening up data to broader groups such as researchers might require institutional review board approval to examine limited data sets. Wide public access has the potential to be exciting—opening up the possibility of a health information economy by allowing anyone to develop innovative analytic measures from the data—but would also raise concerns about such things as the residual identifiability of individual-level data, biased competitive use of the data, and so on. Such broad disclosure
would also allow CMS to share the responsibility to define the “right” derived data with other analysts. This could reduce CMS’s own administrative costs, reduce the cost of data to potential users, and permit the growth of profitable businesses to do useful analyses.
• How are the data organized? Data could be stored centrally within a CMS repository or distributed in some federated manner that keeps the data closer to their source. There are of course tradeoffs involved in whether and how CMS collects and stores detailed clinical data. An indepth discussion of this question is beyond the scope of this report—but such choices have clear implications for CMS’s business and information ecosystems, and so the committee outlines some of these issues briefly.
Benefits to a centralized approach could include:
—The comprehensive ability to measure detailed outcomes;
—Support of large-scale research on comparative effectiveness;
—Public and institutional access to unbiased summary data; and
—Potential for consumers to have direct access to their own integrated data, experience with cases similar to their own, alternative treatments, and so on.
Offsetting these benefits are:
—Technical challenges of operating a national data warehouse for all clinical data;
—Creation of a prime target for security threats; and
—Political challenges related to the role of government with regard to such data.
There are more issues to be sorted out than these, but such tradeoffs will have to be considered carefully.
Other organizational issues include whether the data are left in heterogeneous forms or transformed into a common form on the basis of a consensus set of standards or ontologies, and whether any standards or ontologies extend to the representation of metadata.46 Finally, the desired timeliness of data releases may have to be balanced against the desire for increased utility that will come with transformation into a common format.
• Who is in charge of providing the data? This responsibility could be located within CMS or organized along the lines of a consortium. The responsibility for privacy protections—which may be the largest public
46 An example of the rationale for the use of metadata would be what was recommended by the President’s Council of Advisors on Science and Technology in its December 2010 report to the President, Realizing the Full Potential of Health Information Technology to Improve Healthcare for Americans: The Path Forward, available at http://www.whitehouse.gov/sites/default/files/microsites/ostp/pcast-health-it-report.pdf, last accessed August 8, 2011.
concern—also bears consideration. Another option is a repository, structured similarly to the National Center for Biotechnology Information.47
• What does CMS expect to gain from providing the data? In the short term, CMS’s provision of data may lead to faster, more efficient versions of today’s payment systems, the opportunity for increased scrutiny of the data, and the possibility of innovative analyses by allowing many stakeholders to analyze both their own and others’ experiences as reflected in the data. In the longer term, broader access to data may lead to the construction of better models of clinical outcomes and subsequent improved guidelines for the delivery of high-quality and cost-effective care. It could also foster greater competition among providers and provider organizations on price and quality, by making measures available and by allowing organizations to manage care processes to improve their performance in relation to accepted measures.48
The increasing use of partnerships may allow CMS to see progress in these various arenas of change without being directly and primarily engaged in them. For example, CMS could engage through contracts with consortia of medical and surgical specialty societies that have valuable registries of data and that can, working with CMS, reduce the direct burden on CMS of doing such work alone. These collaborations will also put a premium on data security.
Recent pieces of legislation—including the Medicare Improvements for Patients and Providers Act of 2008, the HITECH provision of the American Reinvestment and Recovery Act, and the Patient Protection and Affordable Care Act—have the potential to improve health care in the United States, and much of that change is data-dependent.
The effective analysis and management of data have the potential to reduce costs, by giving providers the information necessary to choose effective treatments and also by allowing CMS to identify improper payments and prevent fraud; improve overall public health by reducing disparities in treatment and also by rewarding effective outcomes; and empower consumers by providing them with information to manage their own health and also by providing them with information on the quality
47 This model would have an effect on the organization of the data and involve direct interaction with those that fund these capabilities.
48 Other possible transformations are outlined in previous sections of this chapter as well as in Institute of Medicine, 2001, Digital Infrastructure for a Learning Health System: The Foundation for Continuous Improvement in Health and Health Care, Washington, D.C.: National Academy Press.
of providers. A learning health care system should emerge over time, improving the quality, equity, and safety of care for both individuals and populations.
Achieving these goals will be neither easy nor automatic, but with careful attention to the development of a robust, data-driven environment and culture as described in the previous chapters, it is possible.