6

Findings and Recommendations

The National Cancer Policy Board (board) concluded in its April 1999 report, Ensuring Quality Cancer Care, that a cancer data system is needed that can provide quality benchmarks for use by systems of care (e.g., hospitals, provider groups, and managed care systems) (IOM, 1999a). Quality assessment studies would ideally include recently diagnosed individuals with cancer in care settings representative of contemporary practice across the country, using information sources with sufficient detail to allow appropriate comparisons. The board recognized that current data systems and quality assessments were far from this ideal.

This chapter summarizes the board's findings and its recommendations for steps that can be taken to enhance current data systems to bring about sustained improvements in cancer care. The board, in its workshop and deliberations, addressed three questions:

  1. What would the ideal cancer care data system look like?

  2. How are current cancer data systems meeting the needs of healthcare systems?

  3. What steps can be taken to enhance data systems so that they can be used to monitor and improve the quality of cancer care?



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 76
Enhancing DATA SYSTEMS to Improve the Quality of CANCER Care 6 Findings and Recommendations The National Cancer Policy Board (board) concluded in its April 1999 report, Ensuring Quality Cancer Care, that a cancer data system is needed that can provide quality benchmarks for use by systems of care (e.g., hospitals, provider groups, and managed care systems) (IOM, 1999a). Quality assessment studies would ideally include recently diagnosed individuals with cancer in care settings representative of contemporary practice across the country, using information sources with sufficient detail to allow appropriate comparisons. The board recognized that current data systems and quality assessments were far from this ideal. This chapter summarizes the board's findings and its recommendations for steps that can be taken to enhance current data systems to bring about sustained improvements in cancer care. The board, in its workshop and deliberations, addressed three questions: What would the ideal cancer care data system look like? How are current cancer data systems meeting the needs of healthcare systems? What steps can be taken to enhance data systems so that they can be used to monitor and improve the quality of cancer care?

OCR for page 76
Enhancing DATA SYSTEMS to Improve the Quality of CANCER Care WHAT WOULD THE IDEAL CANCER CARE DATA SYSTEM LOOK LIKE? There is no national cancer care data system in the United States. Like the U.S. healthcare system, the data systems available to assess the quality of care on a national or regional basis are fragmented (Pollock, 1997). Advancing quality of care involves applying data in at least three ways: assessing levels and trends in quality of care for whole populations (e.g., the nation, by region, or by state) to identify the magnitude of quality problems and their distribution, determining correlates of quality cancer care (e.g., characteristics of patients and health systems) to elucidate potential causal factors, and measuring and monitoring the quality of cancer care within systems of care to promote quality improvement and allow purchasers and the public to hold systems and providers accountable for the care they deliver. Health services researchers have creatively exploited available databases to meet these objectives, but most sources can be critiqued on one or more important grounds—a lack of geographic representation or the absence of critical data elements needed to adjust results to make comparisons. The board concluded that to meet national quality-of-care objectives, a cancer care data system (which could include several distinct databases) would have the following 10 attributes: A set of well-established quality-of care-measures—a single core set of quality measures must be developed using the best available evidence for the full spectrum of an individual's care—from early detection to palliative and end-of-life care. Reliance on computer-based patient records for information on patient care and outcomes—adoption of information technology can improve the timeliness and accuracy of information on the quality of cancer care. Standard reporting of cancer stage, comorbidity, and processes of care—national quality assessments depend on the uniform recording of data elements needed to accurately assess care. National, population-based case selection—complete ascertainment of incident cancer cases by cancer registries is a prerequisite for national quality assessments, allowing case selection for studies whose results can be generalized to the total population, as well as assessments of quality for important subgroups, for example, individuals of low socioeconomic status, and individuals enrolled in certain types of health plans or delivery systems. Repeated cross-sectional studies to monitor national trends—a series of measures is needed to monitor progress over time.

OCR for page 76
Enhancing DATA SYSTEMS to Improve the Quality of CANCER Care Established benchmarks for quality improvement—systems of care need information on accepted standards of care (e.g., clinical practice guidelines) with which to measure performance. Data systems for internal quality assurance purposes—systems of care need internal data to monitor performance and quality improvement. Public reporting of selected aggregate quality scores—quality measures enable consumers and purchasers to judge the quality of a system of care by its performance relative to evidence-based standards. Adaptability—new evidence on quality measures, changes in healthcare delivery, and technological innovation are among the factors that necessitate flexibility in data systems. Protections to ensure privacy of health information—legal protections and data security systems must be in place to ensure that data collected and stored about an individual's diagnosis and treatment of cancer are used only for legitimate purposes. HOW ARE CURRENT CANCER DATA SYSTEMS MEETING THE NEEDS OF HEALTHCARE SYSTEMS? There are a number of ways for health systems to use available data resources to implement quality improvement programs. Some programs depend entirely on retrospective reviews of medical charts or hospital cancer registry data, while others rely on multiple sources, for example, administrative claims data linked to cancer registry data. Chapter 3 in this report presents 10 case studies to illustrate how various health systems—small physician practices, large integrated delivery systems, professional associations, purchasing coalitions, and states —have used available data to assess cancer care quality (Table 6.1). The case studies are a testament to creativity—data intended for other purposes have in several instances been manipulated to monitor the quality of care and sometimes appear to have been used within programs to effect improvements in care. The data used by systems of care generally fall into one of the following categories: Retrospective medical chart review—abstraction of data from medical charts can summarize performance on selected measures of the process or outcomes of care. Cancer registry data—hospital-based cancer registration programs can provide local data on stage at diagnosis and first course of therapy. In large systems of care, the registry data may be used to identify patients for whom additional clinical information is abstracted from the medical chart. Administrative systems—billing systems and hospital discharge summaries are sometimes used to assess processes of care for identified populations (sometimes through linkages to cancer registry data).

OCR for page 76
Enhancing DATA SYSTEMS to Improve the Quality of CANCER Care TABLE 6.1 Illustrative Case Studies of the Use of Data to Monitor the Quality of Care Name (type of organization) Purpose Data Source(s) 1. Marin Oncology Associates (private oncology practice) Monitor adherence to guidelines on screening, treatment, follow-up, supportive, and end-of-life care Medical chart 2. OnCare (Physician Practice Management Company) Monitor adherence to guidelines on treatment, follow-up, and end-of-life care Electronics medical chart 3. American College of Radiology Monitor patterns of care and adherence to guidelines Medical chart abstraction from a national sample of radiation oncology providers 4. Sutter Health (integrated health care delivery system) Monitor adherence to breast cancer treatment guidelines Hospital cancer registry, State cancer registry, administrative data, medical charts, patient surveys 5. Providence Health Plan (integrated delivery system) Monitor adherence to breast cancer treatment guidelines Hospital cancer registry, administrative data, medical charts, patient surveys 6. National Comprehensive Cancer Network (17 large cancer centers) Monitor adherence to breast cancer treatment guidelines Medical charts. Reporting according to a uniform data set 7. Roswell Park Cancer Institute and private insurers in New York Monitor adherence to breast cancer treatment guidelines Insurance claims, medical charts 8. Colorado Cancer Registry, University of Colorado, and the State Medicare Peer Review Organization Monitor use of adjuvant therapies for breast and colorectal cancer State cancer registry, Medicare claims, medical charts 9. Central Florida Health Care Coalition (business coalition) Monitor quality of care for individuals with selected conditions including cancer Insurance claims, (hospital and outpatient), patient survey 10. National Cancer Data Base (American College of Surgeons, American Cancer Society) Monitor quality of care for individuals with cancer Hospital cancer registries, uniform reporting requirements

OCR for page 76
Enhancing DATA SYSTEMS to Improve the Quality of CANCER Care Patient surveys—patient satisfaction surveys and surveys to assess the quality of life post treatment may be administered among cancer patients. Uniform data collection—clinical information systems with data dictionaries and standardized reporting requirements have been put in place to monitor adherence to guidelines. Prospective medical chart review—computer-based patient records used by a networked group of providers allow the entry and near simultaneous assessment of performance on selected measures of the process or outcomes of care. The case studies demonstrate several barriers to systematic quality monitoring within healthcare systems: A lack of recognized measures of quality—each provider group or organization has taken upon itself the task of reviewing evidence, developing guidelines, and identifying measures. These activities are very costly undertakings, require considerable expertise, and need to be continually reviewed in light of new evidence. A heavy reliance on retrospective medical chart reviews to monitor processes of care—chart abstraction is labor intensive, inefficient, and prone to error relative to the prospective electronic capture of information possible through computer-based patient record systems. An absence of benchmarks with which to measure progress and success—systems sometimes establish internal benchmarks or practice norms, but there is often no way to compare internal performance to that of other providers. Internal benchmarks may be skewed if certain statistical issues are not taken into account (e.g., providers with small numbers of cases can unduly effect norms). A lack of attention to the full spectrum of cancer care—for example, the quality of pain management and end-of-life care may be overlooked. The board did not attempt to survey all cancer-related quality improvement programs but instead wanted to illustrate a variety of approaches. Of note, however, was the difficulty in identifying even 10 initiatives to profile. There do not appear to be many quality improvement programs addressing cancer care, perhaps because of the noted limitations above. On the other hand, innovations within some of the emerging cancer disease management programs and physician practice management companies are noteworthy. Some have developed sophisticated computer-based patient record systems that prompt physicians with guideline recommendations and system-wide practice norms as they provide care to their patients. Such clinical decision support systems can significantly improve the quality of patient care (Classen, 1998; Hunt, 1998). Data are also captured, stored in a central “data warehouse,” and used to monitor adherence to guidelines. The board estimates that approximately 15 to 20% of cancer patients are cared for in environments where these technologies are becoming available (e.g., certain managed care plans and phy

OCR for page 76
Enhancing DATA SYSTEMS to Improve the Quality of CANCER Care sician practice management companies with computer-based patient record systems [CPRs]) (Mighion, 1999). Medical practice lags behind other industries in applying information technologies, but this may be about to change as new internet-based products targeted to healthcare providers emerge (McDonald, 1998). Significant barriers remain to be overcome before there is widespread adoption of new information technologies in the healthcare sector. Uniform standards, for example, do not yet exist to code and format clinical information within computer-based patient record systems. And even though there are techniques to protect the privacy of electronically stored health information (e.g., encryption, password-driven access), public trust in such technologies is lacking (Goldsmith, 2000). The absence of a set of recognized measures of cancer care quality is a clear impediment to quality assessment. The development of guidelines through the National Comprehensive Cancer Network (NCCN) seems to have spurred quality monitoring activities beyond the cancer centers that developed them, suggesting that if a core set of cancer care measures were available, it would be adopted by systems of care for quality improvement programs. Also lacking is a context for measurement. How should health systems focus their quality improvement programs? Should they focus attention on procedures for which there is significant practice variation, consensus guidelines, evidence from randomized clinical trials? A number of patterns-of-care studies have been completed, but with few exceptions, they have not led to vigorous efforts to reduce practice variation. Variation in practice often reflects uncertainty and the lack of good evidence upon which to base treatment decisions. There is, for example, evidence of significant variation in the use of second-, third-, and fourth-line chemotherapy for patients with progressive non-small-cell lung cancer (Smith, 1998). There are no randomized clinical trials comparing best supportive care versus second-line chemotherapy for patients with non-small-cell lung cancer. A guideline could be established based on expert opinion, but guidelines based on sound evidence rather than expert opinion are most likely to succeed in influencing provider practice (OTA, 1994). Priority should be given to implementing measures for which there is practice variation, despite good evidence to support a standard set of practices. For cancer care a few measures would appear to meet this criterion. There is good evidence, for example, to support the use of adjuvant therapy following surgery for breast and colon cancer and evidence of variation in its use (IOM, 1999a). While quality improvement programs launched within hospitals or integrated health networks are well suited to motivating changes in provider behaviors, population-based studies are necessary to assess progress in quality improvement more broadly. Needed are national or regional studies to: assess the extent of quality problems, identify correlates of poor quality care, establish benchmarks, and

OCR for page 76
Enhancing DATA SYSTEMS to Improve the Quality of CANCER Care target interventions to improve care. At this level, data needs shift. Information is needed for entire populations, not just individuals within certain systems of care. Three well-established data sources for assessing care on a national level are available: the National Program of Cancer Registries (NPCR), the Surveillance, Epidemiology and End Results Program (SEER), and the National Cancer Data Base (NCDB). These three national sources of data on cancer have some common elements. They all, for example, contain hospital-reported cases of cancer, but they differ in most other respects: Purpose—While NPCR and SEER are focused on surveillance, NCDB is the only system that was actually designed as a tool for monitoring the quality of cancer care. Consequently, NCDB includes more data elements on treatment than either NPCR or SEER. Geographic coverage—NPCR and NCDB are national in scope, while the SEER registry is limited to a few geographic areas. SEER data are adjusted using population weights to develop national estimates of cancer incidence, but such procedures would not be appropriate in the context of quality-of-care studies because of the unusual distribution of care systems across the country (e.g., managed care penetration, availability of cancer centers). When fully operational, NPCR (together with SEER) has the potential of being a very valuable resource for national cancer care studies because of its near complete geographic coverage. Caseload—Many fewer cases are processed each year by SEER, and resources are focused on ensuring that the quality of data is high. The completeness, timeliness, and quality of data in the NPCR registries have improved markedly in recent years, but significant variation across states remains. Population-based—NPCR and SEER are population-based, and rates of cancer can be derived for their respective covered populations. NCDB is largely limited to hospital-reported cases, and certain patients and types of cancer are known to be underrepresented (e.g., patients diagnosed and cared for in outpatient settings). Also, while facilities approved by the American College of Surgeons' Commission on Cancer (ACoS-CoC) are required to report data to NCDB, nonapproved facilities are not, creating a likely bias in ascertainment. WHAT STEPS CAN BE TAKEN TO ENHANCE DATA SYSTEMS SO THAT THEY CAN BE USED TO MONITOR AND IMPROVE THE QUALITY OF CANCER CARE? The board recommends that steps be taken in three areas to enhance data systems to support improvements in the quality of cancer care:

OCR for page 76
Enhancing DATA SYSTEMS to Improve the Quality of CANCER Care Enhance key elements of the data system infrastructure: quality-of-care measures, cancer registries and databases, data collection technologies, and analytic capacity. Expand support for analyses of quality of cancer care using existing data systems. Monitor the effectiveness of data systems to promote quality improvement within health systems. 1. Enhance Key Elements of the Data System Infrastructure Quality-of-Care Measures Recommendation 1: Develop a single core set of cancer care quality measures. Broad consensus has been reached about how to assess some aspects of quality of care for many common cancers, but specific measures are still being developed and tested within health delivery systems. Quality improvement initiatives for cancer care have been impeded both by the absence of thoroughly tested quality measures and limitations of available data systems. The process of developing and testing such measures needs to be stepped up. As a first step, consensus must be reached on what measures are suitable for immediate use. Major investments in cancer care quality indicators have already been made, but resulting measurement sets have not been widely adopted. The Foundation for Accountability (FAACT), for example, fostered the development of a comprehensive quality indicator set for breast cancer, but few systems of care appear to be using it to assess quality. Barriers to adoption include resistance from plans to new measures, limited resources (e.g., technical assistance), and a lack of technical specifications for adapting measures to different systems of care. Furthermore, given sample size considerations, measures are needed that assess health system competencies across specific diseases, or in the case of cancer, types of cancer (D. Lansky, president, Foundation for Accountability, personal communication, October 13, 1999). There has been a tendency to develop cancer-specific quality measures, but it may be feasible to apply selected measures to all cancer patients or to large segments of patients with cancer. From the board's review of evidence for its report Ensuring Quality Cancer Care there appear to be at least some candidate measures that could be broadly applied that merit systematic review (e.g., documentation of cancer stage in medical chart, use of specific recommended adjuvant therapies, appropriate use of palliation, especially pain control). Since the publication of the board's report, Ensuring Quality Cancer Care, some key initiatives have been launched. Both the National Cancer Institute (NCI) and the American Society of Clinical Oncology (ASCO), for example, have taken steps to identify quality-of-care measures (see below). The National

OCR for page 76
Enhancing DATA SYSTEMS to Improve the Quality of CANCER Care Committee for Quality Assurance (NCQA), an accrediting body for managed care organizations, has designated a medical advisory panel on oncology to identify quality measures. A Department of Health and Human Services (DHHS) committee, the Quality of Cancer Care Committee (QCCC), has been established to focus both on research issues and the delivery of care. The QCCC will work within the structure of the DHHS Quality Improvement Initiative (QII) (NCI, 1999b). While these disparate efforts are laudable, a coordinated public–private effort is needed to achieve consensus on a single core set of cancer care quality measures. Such measures are needed to assess current practice, target interventions, and to monitor improvements in care. Recommendation 1a: The secretary of the Department of Health and Human Services should designate a committee made up of representatives of public institutions (e.g., the DHHS Quality of Cancer Care Committee, state cancer registries, academic institutions) and private groups (e.g., consumer organizations, professional associations, purchasers, health insurers and plans) to: 1) identify a single core set of quality measures that span the full spectrum of an individual's care and are based on the best available evidence; 2) advise other national groups (e.g., National Committee for Quality Assurance, Joint Commission on the Accreditation of Healthcare Organizations, Quality Forum) to adopt the recommended core set of measures; and 3) monitor the progress of ongoing efforts to improve standard reporting of cancer stage and comorbidity. Achieving consensus on a core set of cancer care quality measures may prove to be difficult; however, a number of resources are (or soon will be) available to aid in their identification: cancer care guidelines developed by the NCCN, the NCI's PDQ (Physician Data Query) system, which provides a summary of evidence related to selected cancer treatments, and NCI's forthcoming synthesis of the literature on cancer care outcome measures. Health services research is needed to test promising quality measures within systems of care, and new research initiatives need to be launched within those areas of cancer care for which there is uncertainty regarding best practices. Recommendation 1b: Research sponsors (e.g., Agency for Healthcare Research and Quality [(AHRQ], National Cancer Institute [NCI], Health Care Financing Administration [HCFA], Depart

OCR for page 76
Enhancing DATA SYSTEMS to Improve the Quality of CANCER Care ment of Veterans Affairs [VA]) should invest in studies to identify evidence-based quality indicators across the continuum of cancer care. Wide variation in clinical practice for certain aspects of cancer care exists, which suggests uncertainty among providers (and to some extent consumers) in the face of alternative treatment options for which there are limited data on effectiveness (e.g., use of surgery for prostate cancer, use of second-, third-, and fourth-line chemotherapy for progressive non-small-cell lung cancer). Clinical trials are usually needed to assess the relative effectiveness of various treatment options. Sometimes, the evidence upon which to make practice recommendations exists but has not been systematically examined. In this case, evidence from the literature may be culled to inform treatment practices. AHRQ's “evidence-based practice centers” (EPCs) provide a mechanism to support such evidence syntheses. The EPCs produce science syntheses that provide the foundation for developing practice guidelines, performance measures, and other quality-related activities. Cancer-related topics to date have included assessments of testosterone suppression treatment of prostatic cancer, and evaluation of cervical cytology (www.ahrq.gov/clinic/epc). When evidence-based candidate measures are proposed, they need to be tested within healthcare systems. One model for such applied research is the funding from 1996 to 1998 of cooperative agreements as part of “Expanding and Improving Quality of Care Measures” (Q-SPAN). AHRQ and NCQA cofunded Q-SPAN to develop and test clinical performance measures for specific conditions (e.g., cardiovascular diseases, asthma, hip fractures), patient populations, and healthcare settings (www.ahrq.gov/qual/qspanovr.htm). Similarly, HCFA's Diabetes Quality Indicator Project (DQIP) illustrates another successful model for measurement development. Quality indicators were identified and tested through the coordinated efforts of a group of “technical ” experts (e.g., clinicians specializing in diabetes care, health services researchers) and an “operations” group made up of representatives of the private sector (e.g., NCQA, FAACT, American Diabetes Association, American Association of Family Physicians, American College of Physicians) and public sector (e.g., HCFA, VA, Centers for Disease Control and Prevention [CDC]). The project benefited from a number of other independent research initiatives (e.g., the Medical Outcomes Study, AHCPR's PORT [Patient Outcomes Research Teams], HCFA's Ambulatory Care Quality Improvement Project) and consequently was able to release a fully tested core quality indicator set in 1998, within 1 year of the project 's onset (B. Fleming, Diabetes Quality Indicator Project, HCFA, personal communication, March 24, 2000). In addition to having general indicators of quality, standardized measures of cancer stage and comorbidity are needed so that apparent differences in quality can be correctly attributed to aspects of health care. Differences in reporting requirements that are burdensome for registrars exist now, but efforts are un

OCR for page 76
Enhancing DATA SYSTEMS to Improve the Quality of CANCER Care derway through a collaborative Stage Task Force to standardize data collection for both cancer stage and comorbidity. Recommendation 1c: Ongoing efforts to standardize reporting of cancer stage and comorbidity should receive a high priority and be fully supported. While first steps to identify and adopt cancer care measures are being taken within the cancer care community, a broadly focused response to quality of healthcare issues is taking shape at the federal level. In the wake of the influential report of the President's Advisory Commission on Consumer Protection and Quality in the Health Care Industry (1998) two complementary bodies have been formed on healthcare quality—one in the public sector to promote interagency coordination among DHHS and other federal agencies (Quality Interagency Coordination Task Force [QuIC]), and the other in the private sector to improve healthcare quality, measurement, and reporting (National Forum for Health Care Quality Measurement and Reporting) (see Chapter 5). The aims and activities of both the QuIC and the Quality Forum are quite relevant to the quality cancer care agenda. The QuIC's goal is to ensure that all federal agencies involved in purchasing, providing, studying, or regulating healthcare services are working in a coordinated way toward the common goal of improving quality of care. The complementary Quality Forum established with a private-sector base will focus on measurement issues in an effort to ensure system-wide capacity to evaluate and report on the quality of care and to further consumer understanding and use of quality measures. At the same time, an accountability framework is developing within the private sector that incorporates performance measurement of health plans and other healthcare organizations. Inclusion of cancer care measures into these systems could provide valuable information about cancer care to consumers, purchasers, and providers of care: Health plans report quality-of-care data to NCQA, an accrediting body for managed care plans. Hospitals and healthcare organizations are surveyed and accredited according to standards established by the JCAHO and since 1997 have been required to participate in a quality performance system. In summary, a national public–private organizational infrastructure is in place to focus on quality measurement and improvement. At the same time a “system” of accountability has emerged within the private sector.

OCR for page 76
Enhancing DATA SYSTEMS to Improve the Quality of CANCER Care Recommendation 1d: Efforts to identify quality of cancer care measures should be coordinated with ongoing national efforts regarding quality of care. Cancer Registries and Databases Three national cancer-related databases hold great promise to further quality improvement efforts: NPCR, SEER, and NCDB. NPCR and SEER are cancer surveillance systems with a primary mission of providing population-based estimates with which to understand the occurrence and distribution of cancer. These surveillance systems, however, when linked to other data sources or when used to select individual cases for special studies, have great potential to provide population-based estimates of quality-of-care problems. Although these systems hold promise for such studies, most states do not have the resources to augment their current workload to conduct such studies, which fall outside their primary mission of cancer surveillance. Many states are struggling simply to ensure the basic adequacy of their cancer surveillance systems. The capacity of states to perform cancer surveillance has been greatly enhanced by NPCR. Since 1994, almost all states have received financial and technical assistance from the CDC, and many have adopted model legislation provided by CDC. With this support the registries' ability to ascertain cases has improved, and many now have achieved at least 90% coverage. Many gaps remain, however, and timeliness of reporting is a problem within many registries. The CDC eventually plans to pool NPCR data centrally and to link the data to other sources, for example, Medicare claims and hospital discharge files. The CDC's first call for data will occur in fiscal year 2000. If completed nationwide (and in collaboration with SEER), this would represent a tremendous opportunity to learn more about the quality of cancer care. Relative to its charge (roughly 1 million cases ascertained per year), the NPCR is rather modestly funded (roughly $24 million in federal funding per year in recent years). In addition to what can be learned through linkages, special studies could be conducted among a representative sample of individuals with cancer. The databases themselves lack the detailed treatment and comorbidity information needed for most quality-of-care studies. Through special studies, representative samples of patients can be identified through registries and followed up to collect the more detailed standardized comorbidity and treatment data (e.g., by contacting attending physicians by mail or by abstracting records in oncologists ' offices). NPCR and SEER are best suited to providing information through data linkages and as a frame for special studies, either at the level of the state, or ultimately, when state data are pooled, on a national basis. If the NPCR and SEER data were used in these ways, it would not be necessary for registries to add data elements to collection. They would need to be excellent “incident” registries. The CDC and the NCI SEER programs have a tradition of providing assistance to states and could provide technical assistance to states to facilitate linkage and

OCR for page 76
Enhancing DATA SYSTEMS to Improve the Quality of CANCER Care other special studies. The CDC and NCI have recently agreed to increase their level of collaboration in several areas including the sponsorship of registry-related training activities and conduct of research (DHHS, Public Health Service, “Memorandum of Understanding Between the Centers for Disease Control and Prevention and the National Cancer Institute, ” February 14, 2000, personal communication). Recommendation 2: Congress should increase support to the CDC for NPCR to improve the capacity of states to achieve complete coverage and timely reporting of incident cancer cases. NPCR's primary purpose is cancer surveillance, but NPCR, together with SEER, has great potential to facilitate national, population-based assessments of the quality of cancer care through linkage studies and by serving as a sample frame for special studies. ACoS-CoC and the American Cancer Society (ACS) have long supported the examination of quality of cancer care through the most extensive national data collection effort dedicated to this purpose, NCDB. NCDB has tremendous potential to provide detailed information regarding quality to the approved facilities that report to it (and to the few nonapproved facilities that voluntarily report), thereby encouraging improvements in performance. NCDB is in the process of redesigning its reports to facilities, using a report card format with charts showing facility vs. national norms as benchmarks. Eventually, it may be the case that approval depends in part on performance, or improvement in performance. It is, therefore, an excellent tool for motivating change among providers. As a source of national information on quality, however, NCDB has limitations because of its lack of complete coverage. This coverage problem may worsen as more and more care shifts to outpatient settings. On the other hand, as care shifts to the ambulatory setting, more such facilities may seek commission approval and report to NCDB. Also, NCDB could become a comprehensive source of cancer care data if facilities treating cancer patients were required by JCAHO or other agencies (e.g., HCFA) to be approved. National estimates of quality care could be derived from NCDB, even with its incomplete coverage, if a sample of nonreporting facilities were identified for supplemental data collection. Weighting techniques could then be used to achieve national estimates. Recommendation 3: Private cancer-related organizations should join ACS and ACoS to provide financial support for NCDB. Expanded support would facilitate efforts underway to report quality benchmarks and performance data to institutions providing cancer care.

OCR for page 76
Enhancing DATA SYSTEMS to Improve the Quality of CANCER Care Data Collection Technologies Some of the first “evidence-based” process measures have taxed available data systems. There are two difficulties as far as quality studies are concerned. First, registries only include information on first treatments, and comprehensive quality studies will depend on data related to the full spectrum of cancer care. Second, it is difficult to gather information on treatment that occurs outside of the reporting facility, and retrieval will only worsen as cancer care increasingly moves into outpatient settings. The process measure, use of radiation therapy following breast-conserving surgery for breast cancer, illustrates the difficulties. To assess compliance to this standard, a data system must identify women with breast cancer who have undergone breast conserving surgery, and among this group assess the proportion receiving radiation therapy. Registries may be able to identify the “denominator,” the number of women who underwent surgery, however, information on subsequent use of radiation therapy may be difficult to obtain, especially if the adjuvant therapy took place outside of the facility reporting the case. The shift of cancer care to outpatient settings has exacerbated reporting problems, and analyses of treatment data from cancer registries show substantial underreporting (Bickell, 2000). Despite the growing difficulties of retrieving treatment data, state registries do not receive any direct financial support for data collection from NPCR or SEER. ACoS-CoC has quality of cancer care at the core of its mission. Facilities approved by ACoS-CoC have cancer registrars forward to NCDB the same data reported to the state cancer registry along with some additional diagnostic and treatment-related data. NCDB faces the same problem as state cancer registries of limited access to information on subsequent care occurring outside of the reporting facility. In addition, any case diagnosed and treated outside of a hospital could be missed entirely. The burden to facility-based registrars is great, but like NPCR and SEER, NCDB does not currently provide direct support for data collection activities. How best to collect treatment data for quality-of-care studies needs to be reexamined. Under current systems, state- and facility-based cancer registrars do not have the resources needed to collect complete and accurate treatment data. Perhaps the most effective method of retrieving complete treatment data for quality studies is to link cancer registries to administrative records (e.g., SEER-Medicare, hospitals discharge abstracts) or through special studies. For special studies, registry staff need to have additional resources to gather necessary data. Alternate funding mechanisms for data reporting for quality purposes have been used by purchasers. Gateway Purchasing Association, a purchasing coalition of businesses in St. Louis, for example, withholds 4% of premiums from the health plans with which it contracts, unless the plan provides the quality information requested by the coalition (e.g., HEDIS® measures). Ultimately, such methods would apply to cancer care if cancer-specific indicators were adopted within HEDIS® or other accepted measurement sets.

OCR for page 76
Enhancing DATA SYSTEMS to Improve the Quality of CANCER Care Recommendation 4: Federal research agencies (e.g., NCI, CDC, AHRQ, HCFA) should support research and demonstration projects to identify new mechanisms to organize and finance the collec tion of data for cancer care quality studies. Current data systems tend to be hospital based, while cancer care is shifting to outpa tient settings. New models are needed to capture entire episodes of care, irrespective of the setting of care. Advances in information technology will provide many opportunities to improve the quality of cancer care. Clinicians with access to computer-based patient record and Intranet systems, for example, will be better able to: uniformly apply stage and comorbidity reporting standards while recording patient data, refer to clinical practice guidelines or protocols at the point of care, and quickly transmit formatted data to cancer registrars. There are a few pioneers of such applications among cancer care providers, but adoption of such technologies in health care has lagged behind those of other industries. An Institute of Medicine (IOM) committee charged in 1997 with examining progress toward the development of a CPR concluded that support for CPR research has not been provided in the scope and scale necessary to enable major breakthroughs and that federal funding in the United States has been modest and inconsistent (IOM, 1997). Recommendation 5: Federal research agencies (e.g., National Institutes of Health [NIH], Food and Drug Administration [FDA], CDC, VA) should support public–private partnerships to develop technologies, including CPRs and Intranet-based communication systems, that will improve the availability, quality, and timeliness of clinical data relevant to assessing quality of cancer care. Analytic Capacity Cancer care data, even when enhanced, are of little use if only a few individuals are trained in their analysis and interpretation. Many methodologic issues need to be resolved in establishing quality improvement systems (e.g., setting benchmarks, creating report card formats for providers and consumers, adjusting for case-mix differences), which will require the concerted efforts of clinicians and health services researchers. Several organizations provide training grants and fellowships (e.g., ACS, Robert Wood Johnson Foundation, NCI, AHRQ), and these should be applied to train investigators in these areas. The creation of “Centers of Excellence” in cancer-related health services research could provide focal points for both training and research.

OCR for page 76
Enhancing DATA SYSTEMS to Improve the Quality of CANCER Care Recommendation 6: Federal research agencies (e.g., NCI, AHRQ, VA) should expand support for training in health services research and for training of professionals with expertise in the measurement of quality of care and the implementation and evaluation of interventions designed to improve the quality of care. The board reiterates its recommendation made in Ensuring Quality Cancer Care, that research sponsors should expand support for cancer-related health services research. In response to the NCPB report, NCI has committed to launching a coordinated program of research to improve the methodological and empirical base for quality-of-care assessment for cancer. As conceived, this research program will evaluate whether observed patterns of care are associated with good outcomes, establish the use of a core set of outcome measures in research and medical care applications, investigate methodologic innovations to improve data collection, and, using the above, promote the development of a national cancer data system to monitor the quality of cancer care (NCI, 1999b). NCI also plans to support cooperative agreements with a consortium of investigator teams that might include one or more of the following: academic institutions, cancer registries, professional associations, cancer centers, and other research organizations (NCI, 1999b). ASCO is taking the first steps toward the design of a quality monitoring system for cancer patients. In collaboration with RAND/UCLA and Harvard University, ASCO will explore methods to generate timely, reliable, and valid data regarding the quality of cancer care. ASCO will articulate a set of measures of quality cancer care and design a sampling and data collection system that will provide the needed information to assess quality at the level of the provider (ASCO, 2000). Opportunities for health services research abound with available sources of data. For example, AHRQ's Healthcare Cost and Utilization Project can be used to assess variations in patterns of hospital care and differences in care across systems of care. Only a few cancer-focused health services researchers have analyzed these data. Much more can be learned also from the linked SEER-Medicare database, just updated to include Medicare data through 1998 for persons diagnosed with cancer in 1996. Other recent health services research programs could also provide valuable information related to quality. NCI, for example, has organized a consortium of large managed care plans to promote collaborative research. These health plans have internal information systems and a population-based approach to health care, making them ideal partners for research.

OCR for page 76
Enhancing DATA SYSTEMS to Improve the Quality of CANCER Care 2. Expand Support for Analyses of Quality of Cancer Care Using Existing Data Systems. One issue under debate is whether cancer registries, with their primary mission of cancer surveillance, should be augmented to meet the needs of quality assessment, or whether an entirely new system needs be designed, tailored to the needs of quality assessment. One single, integrated database probably cannot meet all of the various objectives of such systems, for example, cancer surveillance, research, and quality monitoring. Such a system would be complex, cumbersome, and terribly costly. Despite some shortcomings, available data systems have been used in creative ways through linkages and as sampling frames for indepth special studies. With enhancements of elements of current systems, these approaches, if widely applied, could answer many outstanding questions about the quality of cancer care, on a national scale, and without delays of many years between data collection and analysis. One of the most productive strategies researchers have used to assess the quality of cancer care has been to link cancer registry data to either administrative records or hospital discharge files. The data sources are often complementary—cancer registry data have good information on diagnosis and cancer stage but may not record complete information on treatment. Administrative data may lack the diagnostic information but have a good record of treatment encounters. The linkage approach is not without problems. Administrative records may have treatments miscoded, comorbidity data needed to adjust results may be limited, and data elements necessary for complete linkage may be absent. Nevertheless, such linkages have allowed researchers to study variation in cancer care and to make comparisons across systems of care. In one notable example, the appropriateness of treatment of elderly Colorado residents with breast or colon cancer was assessed through a linkage between the cancer registry and the Medicare claims files. Following the finding of significant underuse of adjuvant therapy among older patients (verified by chart review and controlling for comorbidity), an extensive educational campaign was launched to address the problem. This project was undertaken in collaboration with the state Peer Review Organization (PRO). Each state has a PRO funded by HCFA, which evaluates whether care given to Medicare patients is reasonable, necessary, and provided in the most appropriate setting. The most recent contract with HCFA (totaling more than $1 billion) requires PROs to conduct local quality improvement projects on six clinical prioritized areas, one of which is breast cancer (Jencks, 1999). The board recommends that PROs across the country use the Colorado linkage study as a model as they plan their quality improvement initiatives. Many cancer registries are achieving near complete levels of case ascertainment, making them valuable as sampling frames for targeted special studies. Here, cancer registry staff may be asked to gather from medical charts, or contacts with reporting physicians, clinical information to supplement that obtained for registry purposes. With appropriate resources, special studies can be launched relatively quickly in response to specific research questions. Some

OCR for page 76
Enhancing DATA SYSTEMS to Improve the Quality of CANCER Care state laws regarding confidentiality and consent have made timely access to research subjects difficult (e.g., requiring consent from patients and the attending physician); however, this approach provides an efficient mechanism to identify and then conduct follow-up among selected patients. Recommendation 7: Federal research agencies (e.g., NCI, AHRQ, VA) should expand support for health services research, especially studies based on the linkage of cancer registry to administrative data and special studies of cases sampled from cancer registries. Resources should also be made available through the NPCR and SEER programs to provide technical assistance to states to help them expand the capability of using cancer registry data for qual ity improvement initiatives. NPCR should also be supported in its efforts to consolidate state data and link them to national data files. Other opportunities to evaluate cancer care regionally occur without reliance on cancer registry data. For example, a group of private insurers has joined researchers at Roswell Park Cancer Institute to evaluate the quality of care among their covered population using insurance claims data. Coalitions of businesses have also formed to evaluate the quality of care using their pooled claims data. The board applauds efforts by private insurers and employers to measure and improve the quality of cancer within their respective populations. Much can be learned about the quality of cancer care by linking data from cancer registries to other sources, such as insurance records. Questions arise, however, regarding who should have access to the personal identifying information needed to conduct such linkages, and whether the individual identities of patients recorded in electronic databases can be held in confidence. No legal action regarding a breach of confidentiality from cancer registries or other databases is known to have occurred in the United States, but the potential for a breach necessitates that adequate safeguards be in place. Recent federal legislation (the Health Insurance Portability and Accountability Act) established legal sanctions for wrongful disclosure of individually identifiable health information and called for the secretary of Health and Human Services to provide detailed recommendations on privacy of health data and procedures and rules for authorized disclosure of such information (IOM, 1997). As the quality of state cancer registries improves and as efforts to link registry data to other sources proceed, either at the state or national level, models should be developed for how such linkages should be conducted, and how resulting databases can be released to researchers without compromising the identification of patients or providers. Nontechnical approaches to protecting privacy include: limiting researcher access to the data (e.g., requiring a formal application for use), having signed agreements to confidentiality rules, and forbidding publication of analyses at low levels of geography or for very small groups (e.g., rare cancers). NCI follows these and other procedures when releasing to health

OCR for page 76
Enhancing DATA SYSTEMS to Improve the Quality of CANCER Care services researchers data from the SEER-Medicare linked files. The National Center for Health Statistics (NCHS) has established a data center where researchers wishing to link proprietary or other data to national survey data may do so under very controlled conditions (www.cdc.gov/nchs/r&d/rdc.htm). Technical approaches to protecting electronic health information include authentication (e.g., use of IDs and passwords), audit trails (i.e., electronic tracking of access events), and encryption (e.g., limiting access of data to those with an encryption key to decode data) (NRC, 1997). In accordance with the Health Insurance Portability and Accountability Act, DHHS has developed rules setting standards assuring that individual systems have adequate security and an organizational policy to protect data security (Hodge, 1999; Hodge et al., 1999; Ziegler, 1999). Recommendation 8: Federal research agencies (e.g., NCI, AHRQ, HCFA) should develop models for the conduct of linkage studies and the release of confidential data for research purposes that protect the confidentiality and privacy of healthcare information. 3. Monitor the Effectiveness of Data Systems to Promote Quality Improvement Within Health Systems. Ideally, investments in data systems contribute to quality improvement within health systems. Theory would suggest that quality within health systems improves when organizations measure and monitor performance, encourage change through incentive systems and education, and hold providers accountable to the quality expectations of purchasers and consumers. This market-driven approach to quality improvement holds promise, but there are relatively few examples of full implementation and successful outcomes to motivate its widespread adoption. Evidence of the success of data-driven quality improvement initiatives are needed for cancer care. Recommendation 9: Federal research agencies (e.g., NCI, AHRQ, HCFA, VA) should fund demonstration projects to assess the application of quality monitoring programs within healthcare systems and the impact of data-driven changes in the delivery of services on the quality of health care. Findings from the demonstrations should be disseminated widely to consumers, payers, purchasers, and cancer care providers. In summary, the broad availability of cancer-specific data resources makes cancer a unique disease for targeting quality improvement initiatives in patient care. The board is confident that, with a concerted effort, these resources could provide invaluable insights into the quality of contemporary cancer care and point the way to improved care.