The United States has no national cancer care data system. 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 and Rice, 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, by state) or important subgroups (e.g., racial/ethnic groups, the medically uninsured) 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.
Available databases have been creatively exploited to meet these objectives, but most sources can be critiqued on one of two important grounds—a lack of geographic representation, or the absence of critical data elements needed to adjust results to make comparisons. To put the limitations of current data systems in perspective, this chapter describes what might be construed as an ideal cancer care data system. Later, in Chapter 5, the application of current data systems for quality monitoring is assessed in the context of this ideal.
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 11
Enhancing DATA SYSTEMS to Improve the Quality of CANCER Care 2 What Would an Ideal Cancer Care Data System Look Like? The United States has no national cancer care data system. 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 and Rice, 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, by state) or important subgroups (e.g., racial/ethnic groups, the medically uninsured) 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. Available databases have been creatively exploited to meet these objectives, but most sources can be critiqued on one of two important grounds—a lack of geographic representation, or the absence of critical data elements needed to adjust results to make comparisons. To put the limitations of current data systems in perspective, this chapter describes what might be construed as an ideal cancer care data system. Later, in Chapter 5, the application of current data systems for quality monitoring is assessed in the context of this ideal.
OCR for page 11
Enhancing DATA SYSTEMS to Improve the Quality of CANCER Care 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, reliance on computer-based patient records for information on patient care and outcomes, standard reporting of cancer stage, comorbidity, and processes of care, national, population-based case selection, repeated cross-sectional studies to monitor national trends, established benchmarks for quality improvement, data systems for local quality assurance purposes, public reporting of selected aggregate quality scores, adaptability, and protections to ensure privacy of health information. WELL-ESTABLISHED QUALITY-OF-CARE MEASURES At the foundation of an ideal cancer care data system would be a single core set of well-established, “evidence-based” quality measures for the full spectrum of an individual's care—from early detection, to palliation, to end-of-life care. Most measures would be of “processes of care” known through clinical trial research to improve outcomes. Such measures are well suited to quality assessment because if performance falls short, it is clear what needs to be done (i.e., intervene to change process of care). A process measure might identify: overuse of tests or procedures with no known efficacy (e.g., use of bone scans following primary therapy for breast cancer to detect secondary cancer), underuse of tests or procedures known to be effective (e.g., use of radiation therapy following lumpectomy for breast cancer), and misuse of interventions (e.g., too low a dose of chemotherapy). Criteria for evaluating quality measures include: that they are clinically meaningful, scientifically sound, and interpretable as judged by the intended audience (IOM, 1999b; McGlynn, 1998). How robust a particular indicator is can be judged according to the level of evidence available to support the link between a particular process of care and good outcomes (Box 2.1). Common sense might dictate the use of certain measures, despite their lack of evidence regarding effectiveness. Documentation in the medical chart of cancer stage, for example, could be considered an indicator because it is a prerequisite to developing a treatment plan and must be communicated to providers throughout a patient' s care. In addition to meeting standards of evidence (or common sense), measures must be applicable in practice settings. In certain care settings, for example, there may be too few patients available for statistically valid comparisons.
OCR for page 11
Enhancing DATA SYSTEMS to Improve the Quality of CANCER Care BOX 2.1 Levels of Evidence Applied to Clinical Research The “hierarchy of evidence” applied to clinical research (i.e., when the question is whether a given treatment is effective in patients with a specific type of cancer) is well established and agreed upon. The following version is taken from the well-respected U.S. Preventive Services Task Force, proceeding from the most reliable to the least reliable type of evidence (i.e., from grade I to grade III): I Evidence obtained from at least one properly randomized controlled trial. II-1 Evidence obtained from well-designed controlled trials without randomization. II-2 Evidence obtained from well-designed cohort or case-control studies, preferably from more than one center or research group. II-3 Evidence obtained from multiple time series with or without the intervention —dramatic results in uncontrolled experiments (e.g., the results of the introduction of penicillin treatment in the 1940s) could also be regarded as this type of evidence. III Opinions of respected authorities, based on clinical experience, descriptive studies and case reports, or reports of expert committees. SOURCE: U.S. Department of Health and Human Services, Office of PublicHealth and Science, 1996, p. 862. An ideal cancer care data system would also provide information on the healthcare experience of individuals with cancer. Optimally, individuals within the care system would report that their care had been well coordinated, that they had easy communication with their providers, and that they felt their care had been consistent with their personal preferences. The quality of an individual's experience within the cancer care system—whether care was perceived to be well coordinated, respectful, supportive, and compassionate—would be assessed through validated survey instruments. Measures for which there is suspected variation, or low overall performance, would be selected to assess quality. If adherence to a standard were known to be uniformly high, there would be no good reason to monitor it. And if quality had improved to meet or exceed a target, that measure might be dropped from the indicator set. The measurement set would be a dynamic one, with additional process measures being added as they are identified through research, and old ones dropped as national (or regional) norms reach established targets. Certain key measures could be maintained to allow for analyses of trends.
OCR for page 11
Enhancing DATA SYSTEMS to Improve the Quality of CANCER Care COMPUTER-BASED PATIENT RECORDS In an ideal system, healthcare providers could easily record patient care data using a computer-based patient record (CPR) system. The entry system would be “smart” and prompt providers to adhere to standards for reporting stage, comorbidity, processes of care, and care outcomes (e.g., complications, indicators of quality of life). CPR systems have the capacity to transform patient care and improve quality. Benefits of CPR systems include (IOM, 1997): integrated view of patient data: patient data are accessible whenever and wherever clinical decisions are made, independent of where the data was originally acquired; access to knowledge resources: providers can access medical and administrative knowledge at the time decisions are made; physician order entry and clinician data entry: systems allow proactive influence on physicians' practice patterns; integrated communications support: activities of healthcare professionals from multiple organizations at different sites can be coordinated; clinical decision support: prompts regarding clinical guidelines, drug interactions, and abnormal laboratory results can improve the clinician's efficiency and compliance with accepted standards of practice. With CPRs, healthcare providers would have access to information at the point of care to aid in clinical decision making and patient counseling. For purposes of cancer registration, CPRs could automate the abstraction of necessary data and dramatically improve the timeliness of reporting. Intranets, controlled-access versions of the Internet, could be set up to facilitate data exchange between clinicians and registries. STANDARD REPORTING Outcomes of treatment for cancer vary markedly by stage of illness (a measure of how advanced cancer is) and by the degree to which patients have other diseases or illnesses along with their cancer (called comorbidity). When making comparisons between groups of patients (e.g., comparing surgical outcomes among patients cared for in large versus small hospitals), it is essential to control both for their stage of illness and comorbidity. Without such controls, worse outcomes could be attributed to differences in care, when in reality they are due to differences in the mix of patients in the two types of hospitals. Quality assessments depend on the accurate recording of stage and degree of comorbidity because what is considered appropriate treatment varies by these patient attributes. An ideal cancer care data system would include information on cancer stage and comorbidity, reported in a standard way.
OCR for page 11
Enhancing DATA SYSTEMS to Improve the Quality of CANCER Care NATIONAL, POPULATION-BASED CASE SELECTION Convenience samples are frequently used for quality studies, but the results of such studies are often difficult to interpret because of their limited population coverage. A study of the quality of cancer care conducted in a few states or within a particular health plan, for example, might highlight problems in quality; however, broader inferences from such studies to the care received in other areas or other plans are difficult to make because the population from which the results are drawn usually differs in important ways from the broader population. The determinants of quality of care have not yet been well established, but evidence suggests the presence of significant geographic variation in patterns of cancer care that persist even after adjustments are made for characteristics of patients and their access to services (Schuster, 1998a). There is much interest in how aspects of healthcare delivery affect the quality of care, and certainly the organization of health care varies markedly by geography. Managed care penetration, for example, is very high in certain states (e.g., California, Minnesota) but extremely low in some states in the South (Modern Healthcare, 1999). Other evidence suggests that certain groups of patients are more prone to poor quality care, for example, individuals of low socioeconomic status and those who lack health insurance coverage. If a quality-of-care study relied on data from areas that differed in sociodemographic make-up from the nation as a whole, results may not accurately reflect the state of quality of care for the nation. In an ideal study of cancer care quality, each newly identified cancer patient in the United States would have a chance to be in the study, and the probability of inclusion would be known. While this approach sounds simple, it is quite difficult to achieve. It requires having a complete listing of individuals with cancer for a defined geographic area. Because cancer treatment differs so markedly by type and stage of cancer, quality studies are often targeted to specific types and stages of cancer. Some cancers are extremely rare, and complete case ascertainment may be needed to capture an accurate assessment of quality of care. For common cancers, however, careful sampling techniques may be applied to obtain a representative group of patients for study. Ideally, each state would have accurate, timely reports from all cancer care providers in the state (and from providers out of state who diagnosed a patient residing in that state) so that lists or “frames” could be developed for sampling purposes. Studies designed to be representative of a total population are often referred to as “population-based ” studies. Many concerns about quality care relate to the initial stages of care—diagnosis, treatment, and follow-up care. Studies of these phases of care can rely on samples from among the 1.2 million new cases of cancer expected annually. Other means of patient selection could be used to study the quality of care at the end of life. For this phase of care, prospective studies might be conducted among a cohort of newly diagnosed patients with cancers having high associated mortality, or with samples of seriously ill patients identified in hospitals, nursing homes, or hospices.
OCR for page 11
Enhancing DATA SYSTEMS to Improve the Quality of CANCER Care The selection of patients for study need not occur at the level of the individual with cancer. If a comprehensive listing of cancer care providers were available, a multistage process of case identification could be used. First, a sample of providers could be selected, then within practices, patients (or a sample of patients) could be selected who met entry criteria. REPEATED STUDIES TO MONITOR NATIONAL TRENDS National studies aimed at assessing the quality of cancer care would ideally be repeated at regular intervals to measure progress toward improvement goals. Just as national surveys are conducted regularly of the general population, healthcare providers, and certain healthcare institutions to monitor the achievement of health objectives for the nation (e.g., Healthy People 2010), so too should there be assessments of the quality of care. Trend data may convey the feasibility of reaching established goals, and variation in rates of change across regions can establish what is achievable within a given space of time. ESTABLISHED BENCHMARKS FOR QUALITY IMPROVEMENT If information on processes of care were available on a national sample of recently diagnosed cancer cases, contemporary patterns of cancer care could be described. These national patterns of care data could then be used to assess compliance to accepted standards of care and to establish specific benchmarks, or targets, for the improvement of care. The benchmarks would be set in such a way that they represented excellence, and at the same time would be achievable by practitioners. DATA SYSTEMS FOR LOCAL QUALITY ASSURANCE PURPOSES Opportunities to change practice behavior and improve the quality of cancer care rest within local systems of care, for example, hospitals, health plans, and provider groups. The degree to which data can be provided at local levels rests in part on the cancer caseload. For statistically valid comparisons to be made, a sufficient number of individuals with cancer must be present in each site of care; however, data from smaller units can often be aggregated into larger service areas. Alternatively, it is sometimes possible to apply general measures of quality across discrete patient populations. It may be feasible, for example, to apply a general indicator for appropriate use of adjuvant therapy across several types of cancer (e.g., breast, colorectal) or appropriate use of palliative care (e.g., pain management) among individuals with advanced or recurrent cancer.
OCR for page 11
Enhancing DATA SYSTEMS to Improve the Quality of CANCER Care PUBLIC REPORTING OF SELECTED AGGREGATE QUALITY SCORES An ideal cancer care data system would allow hospitals and health plans to assess their care relative to national or regional norms and to identify ways that care could be improved. Facilities and plans would receive periodic, easy-to-read charts comparing their recent experience against national or regional norms. For health plans, performance scores could be published by national accrediting bodies such as the National Committee for Quality Assurance (NCQA). Hospitals could have scores considered by groups such as the Joint Commission on Accreditation of Healthcare Organizations (JCAHO). Publicly available information on quality could potentially inform decisions about care made by consumers and healthcare purchasers. ADAPTABILITY Even though consistency in measurement is often desirable so that trends can be accurately monitored, cancer care data systems need to be flexible so that accommodations can be made for new evidence on quality measures, changes in healthcare delivery, and technological innovation. PROTECTIONS TO ENSURE PRIVACY OF HEALTH INFORMATION Maintenance of sensitive personal health information, such as the diagnosis of cancer, in large computerized databases raises serious issues regarding privacy and confidentiality. Legal protections must be in place to ensure that data collection is appropriate, that information is stored securely, and that access to the information is controlled. Federal and state laws and regulations governing privacy, confidentiality, and data security must be strictly enforced while at the same time allowing important registry functions to proceed. Data linkages using personal identifiers such as social security number or birth date, for example, are necessary to eliminate duplicate reports of a case from different healthcare providers. A relatively new application of linkage is the assessment of quality of cancer care.