|
Advertisement |
|
|
||
 |
|
|||||
|
|||||
|
 |
||||||
|
||||||
|
||||||
|
© 2002 American Society for Clinical Oncology The Case Identification Challenge in Measuring Quality of Cancer CareByFrom RAND Health and RAND Statistics Group, RAND, Santa Monica; University of California Los Angeles (UCLA) Schools of Medicine and Public Health, and Department of Medicine, UCLA Center for Health Sciences and Division of General Internal Medicine and Health Services Research, Los Angeles, CA; and Merck-Medco Managed Care, LLC, Franklin Lakes, NJ. Address reprint requests to Katherine L. Kahn, MD, RAND, PO Box 2138, 1700 Main St, Santa Monica, CA 90407-2138; email: kkahn{at}rand.org
PURPOSE: The delivery of quality care to all patients with cancer has been named as a national priority within the American health care system. This article addresses the issues critical to case identification in cancer quality measurement and recommends possible strategies for accurately identifying a population of cancer patients. METHODS: We present the measurement issues associated with the basic challenges of case identification strategies for quality measurement. We discuss two basic challenges: (1) accurately identifying all patients with the defining characteristics (eg, a diagnosis of breast cancer), and (2) identifying only patients with these characteristics. RESULTS: Possible options for identifying newly diagnosed patients include using claims or other administrative data, cancer registries, cancer registry rapid case ascertainment, pathology laboratories, and physicians’ offices. In the published literature, the sensitivity of claims varies from 75% to 95%, whereas central registries must have a 90% completeness rate to be certified. Most of these approaches, however, involve limitations to obtaining valid and comparable data across multiple settings. CONCLUSION: Using an existing data collection system staffed by skilled data collectors and managers should result in substantially more accurate and timely data. Registry officials and the government agencies that provide their support should be encouraged to adopt quality-of-care analyses as an important purpose of the registry system and to enhance their capacity to rapidly ascertain cases, collect the appropriate identifying information needed for patient contact, and verify stage at diagnosis. In order to meet the growing demand for timely, accurate information about quality of care, registries are likely to require additional support so they can enhance their capacity to rapidly ascertain cases, collect the appropriate identifying information needed for patient contact, and verify stage at diagnosis.
MAJOR PURCHASERS, consumers, and providers are pressing for better information on the quality of health services. Although much of the early focus on quality measurement in health care has been on preventive services,1 stakeholders are becoming increasingly interested in measuring the quality of care provided for patients with serious illnesses, such as cancer. The American Cancer Society recently held a national conference, "Purchasing Oncology Services: Current Methods and Models in the Marketplace," to address growing concern that the quality of cancer care should be of primary interest to health care purchasers.2 The Institute of Medicine’s National Cancer Policy Board recently focused the nation’s attention on the quality of cancer care when they concluded that "for many Americans with cancer, there is a wide gulf between what could be construed as the ideal and the reality of their experience with cancer care."3 Although purchasers, consumers, and providers are now demanding accurate and timely information about the quality of cancer care, many methodologic challenges must be overcome if we are to meet their demands. Quality of care is "the degree to which health services for individuals and populations increase the likelihood of desired health outcomes and are consistent with current professional knowledge."4 Quality measures may assess the processes of care, what physicians and nurses do to patients, or the outcomes of the care experienced by the patient.5,6 Measures should be supported by scientifically sound research demonstrating that better performance of the processes is associated with improved outcomes for patients. Quality can be measured at the level of the individual (eg, physician) or the organization (eg, health plan, hospital, medical group) and then compared across these entities or to a benchmark. However, obtaining comparable information across different organizations is particularly challenging in today’s heterogeneous and changing world of health care delivery. Different organizations frequently use quite different information systems, and these fundamental differences in the availability, reliability, and validity of data sources make the collection of comparable data difficult. Nevertheless, the information must be comparable across all units of comparison for the stakeholders—consumers, providers, and purchasers—to have confidence in the results of quality measurement. A number of studies have examined patterns of cancer care, primarily for breast cancer patients.7-20 Unfortunately, data from these studies are limited from a quality measurement perspective because it cannot usually be determined whether the patients who did not receive the treatment in question were the recipients of substandard care or there was another acceptable reason for the patients not getting treated. In other words, the measures used in patterns of care studies are not defined precisely enough to be able to state with confidence whether the care provided was of good or substandard quality. Quality measures have two distinct components: the denominator and the numerator. The denominator defines the population of interest, that is, all patients who are eligible to receive a particular health care process or a given outcome. The numerator defines the proportion of eligible population who actually did (or did not) experience the process or outcome. The data collection steps necessary to construct the denominators are distinct from those necessary to construct the numerators. The unique data collection requirements for the numerator and denominator of a quality measure each present methodologic challenges that threaten the reliability and validity of the measure. The denominator is particularly sensitive to such threats because it is dependent on the strategy chosen for case identification. The purpose of this article is to clarify the multiple issues critical to case identification in quality measurement for cancer care and to recommend possible strategies for handling these issues. Ways in which research findings might be altered as a function of the case identification strategy are discussed.
Design: The Challenge of Cancer Case Identification in Quality-of-Care Measurement A case identification strategy for quality measures presents two basic challenges: (1) to identify accurately all patients with the defining characteristics (eg, a diagnosis of breast cancer), and (2) to identify only patients with these characteristics. Biases can result if the case identification strategy is only capable of identifying a convenient-to-gather but nonrepresentative portion of cases.21 For example, if the case identification strategy does better at identifying highly educated patients, the performance measurement results might present an inaccurate picture of the care provided if better educated patients tended to receive better care. If the strategy identifies patients using certain diagnostic procedures (eg, those reimbursed through claims), the performance assessment would miss those who did not experience one of these procedures. Not identifying cases because they did not undergo the procedure would be especially problematic if the procedure was requisite for good quality care. Serious ethical concerns are raised, conversely, if patients without cancer are identified as cases and then contacted as if they have cancer. Case identification for quality assessment of cancer care presents several specific challenges. First, most of the focus in quality assessment is on incident cancer cases, so researchers need a case-identification strategy that allows them to separate incident from prevalent cases. Second, disease severity, defined as the stage of cancer, is often also important for cancer quality measurement because researchers or policy-makers may be interested in the care of a specific subset of patients (eg, patients with early breast cancer). Third, a cancer diagnosis is generally considered sensitive information, creating special challenges for protecting the rights of human subjects. Although these issues may also be important in quality measurement in other clinical conditions, overcoming these challenges is a necessary prerequisite for the evaluation of the quality of cancer care. A number of potential strategies exist for identifying patients with cancer. Possibilities include using claims or other administrative data, cancer registries, cancer registry rapid case ascertainment (RCA), pathology laboratories, and physicians’ offices. The validity and reliability of these methods has been improving with time as billing has increasingly been tied to claims and as informatics systems have improved. The section that follows describes these strategies and the potential capacities and limitations they offer for accurately identifying all new cancer cases consistently for measurement of different health care services in terms of data accuracy, timeliness, issues of subject consent, confidentiality, the burden on clinicians, and resource requirements. Table 1 lists the relative advantages and disadvantages of these different strategies.
Claims Data Claims and other administrative databases can be searched for diagnoses of cancer and procedures related to cancer. However, the reported sensitivity of claims for identifying incident cancer cases ranges from 70% to 95% when compared with cancer registry data.22-27 Moreover, not all cases identified by diagnostic and procedure codes in claims data are valid breast cancer diagnoses. Using medical record data to evaluate a claims-based algorithm for identifying breast cancer patients in a health maintenance organization setting, one study found that 17% of the patients identified through the claims data were incorrectly identified because of coding errors, diagnostic errors, and up-coding.28 One half of those incorrectly identified involved negative biopsies. This study did not evaluate the number of breast cancer patients not identified by the claims data. A study by Warren et al24 comparing Medicare claims with the Surveillance, Epidemiology, and End-Results (SEER) cancer registry found that the positive predictive value of a breast cancer diagnosis on any Medicare claim was only 36.3%: more false-positives were identified than true breast cancer cases. The positive predictive value of claims for identifying cancer cases can be improved with algorithms that combine procedures. However, when attempting to identify incident breast cancer cases, increasing the specificity of claims by using more restrictive rules is associated with a corresponding loss in sensitivity.26,29 In addition, administrative data generally provide no information about the cancer stage at diagnosis and, unfortunately, even when available, claims data on stage do not appear reliable. Although claims data present challenges in identifying patients from fee-for-service settings, there are potentially even greater limitations to their use in identifying cancer patients from HMOs. Some HMOs do not collect any claims or they may be limited to services that the HMO contracts to specialist providers or hospitals. Even if physicians are not required to submit claims for payment, many HMOs require physicians to submit a record of each encounter with a patient that contains his or her identifying information and diagnosis. However, although a health plan may mandate the reporting of such encounter data, in capitated systems, which do not have the incentive of receiving payment for providing the encounter information, such data are likely to be incomplete and subject to selection bias. Omitted cases could in fact be those where the poorest quality care was provided.30 The timeliness of claims data has not been reported in the literature. The studies cited above that have used claims data have publication dates 6 to 8 years after the dates that the services were rendered. Theoretically, claims should be available within months after a service is rendered; however, this is likely to vary substantially across plans and be highly dependent on their degree of infrastructure and the sophistication of their information systems. When claims data are used to identify human subjects according to a specific diagnosis, either to contact or to link their claims data with other data sources, the participant’s consent is required. Although many health plans obtain consent from their enrollees to use their personal information for quality assessment and improvement activities, increasing concern over individuals’ privacy may limit such blanket consents.31,32 In summary, although claims data are often easily accessible and cost little to obtain, problems with the sensitivity and positive predictive value of these data for identifying incident cancer cases limit their usefulness in quality measurement.
Cancer Registries Many hospitals maintain registries on cancer patients under their care, reporting either for American College of Surgeons certification or to meet state or federal requirements. Whereas hospital registries are oriented towards administrative and patient care purposes,34 regional and state-level population-based registries are oriented toward research and planning.35 Some focus on cancer incidence determination as their primary goal; others are oriented toward epidemiologic research or cancer control.36 Reporting requirements are not uniform among registries, although there is an increasing focus on standardized or common data items.37 Many hospital registries report to standards of the American College of Surgeons’ NCDB, whereas many central registries use North American Association of Central Registries data requirements, and the National Cancer Institute (NCI) has specified SEER data requirements.37 Even though many hospitals have cancer registries, hospital registries cannot be solely relied on for case identification because of the increasing proportion of both diagnosis and treatment in the ambulatory setting.38 In addition, use of hospital registries for case identification necessitates decentralized data collection, a major endeavor for any research team. To illustrate, the American Hospital Association Guide to the Health Care Field lists 39 hospitals in Maricopa County in Arizona, 14 of which have cancer programs approved by the American College of Surgeons and therefore likely to have a cancer registry.39 Thirty-seven hospitals are listed for Dade County, Florida, eight of which have cancer programs approved by the American College of Surgeons. After the hospitals with registries are identified, each hospital would need to be contacted and agree to release their data. In addition, approval from each hospital’s human subjects protection committee would be required. Finally, researchers would need to be sensitive to the possibility that working directly with individual hospital registries might be viewed by regional or state registries as delaying their operations. Centralized registries, by definition, obviate the case identification problems related to decentralization. However, it is important to note that all of the centralized registries rely on the hospital registries reporting to them to identify the majority of cancer cases. Thus, any problems with the quality of data at the hospital-level registries will also be reflected in the centralized data systems. Not all central registries are population-based. Reporting to many state registries is conducted on a voluntary basis. For example, the Virginia Cancer Registry includes only those cases diagnosed at approximately 50 hospitals, representing 85% of hospital beds in Virginia.23 The American College of Surgeons NCDB is the largest cancer registry in the United States; however, it also relies on voluntary reporting. In 1994, the NCDB received data from 1,227 hospitals (representing 57% of the estimated cases in the United States).40,41 The NCDB is expected to collect data on 80% of cancer cases diagnosed in the United States in the near future. Population-based registries, which include many of the state cancer registries and the SEER registry, by definition, have the advantage of inclusiveness. A study of reporting accuracy in six SEER regional registries found a 97% completeness rate.42 To be certified, central cancer registries must have a 90% completeness rate. Before 2000, SEER collected data from population-based regional cancer registries in Connecticut, Atlanta, rural Georgia, Detroit, Iowa, Arizona, Hawaii, Los Angeles, New Mexico, San Francisco-Oakland, San Jose-Monterey, Seattle-Puget Sound, and Utah, and included approximately 14% of the population of the United States.43 In 2000, SEER added registries in Kentucky, Louisiana, and New Jersey and expanded representation in California. With this expansion, SEER increased its representation to 26% of the total population of the United States.44 All SEER registries have similar reporting requirements, and the data are subject to verification procedures both at the NCI and at each registry.43 In 1992, Congress passed the Cancer Registries Amendment Act, creating the NPCR, administered by the Centers for Disease Control and Prevention (CDC) to organize the state cancer registries into a national system.45 Since 1994, through the NPCR, the CDC has developed model legislation and regulations to enhance the viability of registry operations; helped states without registries to plan and establish registries; trained registry personnel; improved existing cancer registries so that they meet standards for data completeness, timeliness, and quality; and established a computerized reporting and data processing system.46 In 1999, the CDC provided $24 million of support to cancer registries participating in NPCR to enhance their data collection activities.47 Before NPCR, 40 states had cancer registries, but not all were population-based or able to gather complete and accurate data.47 Most of these state cancer registries lacked adequate resources for ensuring minimum standards for quality and for completeness of case information. For example, in 1990, the Virginia Cancer Registry included cases diagnosed at approximately 50 hospitals, representing 85% of hospital beds in Virginia.48 By 2000, all 50 states and the District of Columbia had population-based registries49; 46 are part of the NPCR and four are sponsored by SEER. In 2000, the North American Association of Central Cancer Registries certified 39 of the state cancer registries (on the basis of their reporting of 1998 cases).50 The main current limitation of SEER, the state registries, and the NCDB, is the time delay from diagnosis to when the data are available, usually 2 years (W. Wright, personal communication, February 1997).43 Although this limits the utility of registry data for prospective studies, retrospective studies of cancer care have been able to use registry data for case identification.51 The requirements to obtain the requisite institutional review boards’ (IRBs) approvals vary significantly with the centralization of the registry. Working with any of these case identification strategies will involve review by at least one IRB. However, using the decentralized approach of identifying cases from individual hospital registries, or using the NCDB (which does not centrally maintain information that could be used to identify patients), would require gaining the approval of each of the participating hospitals’ IRBs, necessitating both a significant amount of time and researcher effort. Because the NCDB does not contain identifying information, researchers who use the NCDB for case ascertainment must work with the individual hospital registries to obtain the contact information and undergo IRB review at each participating hospital. This approach is currently being used by the American Society of Clinical Oncology in a study of breast and colorectal cancer in five cities.52 In addition, most registries require notification of the physician who performed the biopsy, if patients are to be contacted by the researchers.53 Physician notification allows an exclusion screen where physician input advises against contacting a patient who might be harmed by such contact (eg, a patient with psychiatric problems). In some sites, the registry itself may want to contact the physician first.54 Researchers may also want to contact the physician to verify information obtained from the registry, such as patients’ addresses or diagnoses, or to obtain patients’ phone numbers if they are not available through the registry. The timing, efficacy, and cost of these additional services need to be considered in planning a case identification strategy.
Cancer Registry RCA Reporting The specific data items collected through RCA reporting vary, but many include limited demographic information and identification of the patient and the physician who requested the pathology report. These data are sufficient for identifying newly diagnosed cancer patients and linking registry data to other data files, but they do not include information on stage at diagnosis or treatment. The fact that stage at diagnosis is not assigned as part of the rapid reporting means that any prospective study cannot be restricted to selected stages without first getting additional information (ie, contacting the patient’s physician or reviewing medical record data) to obtain the information about each patient’s stage. Many RCA systems do not routinely include the patient’s phone number, but may be willing to negotiate collecting it. If not, the phone number will have to be obtained from the physician’s office if quality-of-care data are to be obtained by surveying the patient by telephone. Registries with RCA reporting are thus a promising source of early case identification data. For example, Potosky et al57 have developed an extensive set of methodologic, clinical, and policy-relevant studies on the basis of a cohort of patients identified promptly after diagnosis with RCA. In sites without formal RCA systems, it may be possible to obtain registry cooperation in reporting similar data quickly. Many registries may have the capability to implement RCA procedures if provided with the necessary resources.
Pathology Laboratories and Physicians’ Offices An alternative approach that is limited to breast cancer case identification would be to review mammography reports. Breast cancer cases would then need to be identified from among the patients with abnormal mammograms. (Collecting case identification data in the radiologist’s office also offers the opportunity to collect baseline quality of care and functional status data from all women having mammograms. Although not germane to case identification per se, a serious problem for outcome measurement is the lack of a premorbid [i.e., before breast cancer] set of data about baseline function, quality of life, and comorbidity. Defining the study population according to those receiving mammogram screening would allow patient survey at the time of mammography and allow collection of premorbid data.) Use of mammography reports would allow identification of two additional patient cohorts, those with no abnormality noted on the mammogram, and those with a possible abnormality or indication on the mammogram of the need for further diagnostic studies, which would facilitate performance measurement in the areas of screening and diagnostic care in addition to breast cancer care. This method of case identification is also extremely decentralized, with the problems associated with decentralization discussed above with respect to using hospital registries or pathology laboratories; however, there are at least two additional problems with this approach. First, only about one in 100 women receiving a mammogram will have breast cancer. So clearly the strategy will be cost effective only if the data about the women who constitute the denominator (ie, all women who receive mammograms) will contribute data relevant to additional analyses. A second problem is that approximately 10% to 15% of women are diagnosed with breast cancer in the setting of a normal mammogram.59 The proposed strategy would not identify these cases unless the women with normal mammograms were identified by another mechanism. Finally, researchers could theoretically collect case identification information directly from the diagnosing physician. The physician might be asked to complete and submit a one-page form at the time the pathology report was received with the breast cancer diagnosis. This approach is limited, however, as it shifts a considerable portion of the burden directly to the physician and is subject to serious threats to validity resulting from expected reporting bias where cases representing compromised quality of care might be differentially omitted.60
Most of the approaches to case identification involve some serious limitations to obtaining valid and comparable data across multiple settings. The problems with accuracy of case identification make claims data alone unacceptable for case identification in most, but not all, settings. Given that burdensome and expensive primary data collection efforts are required with the decentralized approaches of identifying cases through pathology laboratories, radiology offices, or physicians, cancer registries generally offer the best method of identifying incident cancer cases. Thus, cancer registries, by offering the advantages of accurately identifying incident cancer cases, and also providing additional information, such as the stage of disease, initial treatment, and the contact information for some of the physicians treating the patient, represent the best strategy for identifying incident cancer cases. The main limitations to using standard registry data for case identification include the 2-year time lag before data are available to researchers, and the fact that population-based cancer registries represent only certain regions of the country and the largest cancer registry,41 the American College of Surgeons NCDB, is a convenience sample of hospital registries.40 Many researchers have attempted to overcome the problems associated with the poor positive predictive value of administrative data for cancer diagnoses by linking claims data with cancer registry data.61 Although they are successfully able to limit their sample to only those incident cases of cancer in this way, this method still does not overcome the problem of potentially not identifying all of the cancer cases in the population of interest. In addition, linking two separate databases to identify cases requires not only having access to both data sources, but also having administrative or legislative exemptions from having to obtain permission from potential subjects, because performing such a linkage requires knowledge of both the individuals’ identities and their cancer diagnoses. Timeliness is another critical factor in selecting the optimal case identification strategy. The aforementioned data sources frequently present time constraints that become important if quality measures are time sensitive, as are patient self-report measures, or if the results are intended for use in ongoing quality improvement. Quality measurement frequently relies on patient self-report of outcomes (eg, quality of life, patient satisfaction), processes of care (eg, preference communications, procedures), and comorbid conditions. In such cases, the research design needs to maximize the patient’s accurate recall of these outcomes, processes, and conditions, so the data need to be collected soon after the event of interest.62,63 Quality measures intended for use by providers in quality improvement activities also have demanding time frames. Because feedback of information on current practice is critical, it is again necessary to identify cases as soon after diagnosis as possible. When patient self-report data are to be collected soon after diagnosis, the RCA approach appears to have the least limitations. Although the decentralized approaches of pathology laboratories, radiology offices, or physician offices could also potentially be used to identify patients shortly after diagnosis, these three approaches represent new primary data collection efforts that are costly, burdensome to the personnel at the sites, and would require a major effort before data collection could begin to ensure that all IRBs with potential jurisdiction had approved the project. The RCA approach, in contrast, relies on an existing data collection system with familiar requirements and monitoring procedures already in place. Using this existing infrastructure staffed by skilled data collectors and managers is likely to result in substantially better data in terms of accuracy, response rate, and completeness. For these operational reasons, the RCA strategy, when available, is a desirable method for identifying incident cancer cases within a few months of their diagnosis. Currently, any attempt to collect timely data on the quality of cancer care on a national sample would require negotiating RCA on a site-by-site basis. A strategy that relies on a data source that varies from site to site is neither optimal nor sufficient in the long run. Registry officials and the government agencies that provide their support should be encouraged to adopt quality-of-care analyses as an important purpose of the registry system. Because much quality-of-care analysis today relies on patient self-report, and any quality improvement efforts depends on timely data, registries should be encouraged to enhance their capacity to rapidly identify cases and collect the appropriate identifying information needed for patient contact when projects needing these methods are required.64 In addition, more rapid verification of stage at diagnosis is of value. Identifying a population-based cohort of incident cancer cases is extremely difficult with any of the other usual data sources for quality-of-care measurement. Because of their regulatory authority, cancer registries are uniquely situated to identify a population-based sample of cancer patients and remain the best candidate to provide the infrastructure for measuring the quality of cancer care. In order to meet the growing demand for timely, accurate information about the quality of cancer care, registries should be provided with additional support so that they can enhance their capacity to rapidly ascertain cases, collect the appropriate identifying information needed for patient contact, and verify stage at diagnosis.65,66 Case identification is an important and complex undertaking. An easy or standardized approach for identifying cancer cases for quality measurement has not yet been worked out, but cancer registries have considerable potential for accurately producing the information needed in a timely manner. Quality measurement researchers would be well advised to work closely with cancer registry officials to realize this potential.
Supported by contract no. 500-95-0056, task order no. 3, issued by the Health Care Financing Administration of the United States Department of Health and Human Services, Baltimore, MD; contract no. 1-R01-CA81338-01A1 (K.L.K.) issued by the National Cancer Institute, Bethesda, MD; and the Agency for Health Care Research and Quality, Rockville, MD. We thank the staff assistance of Danielle Letourneau and Wendy Parker.
1. National Committee for Quality Assurance: HEDIS 3.0: Narrative—What’s In It and Why It Matters, vol 1. Washington DC, National Committee for Quality Assurance, 1997 2. Cangialose CB, Blair AE, Borchardt JS, et al: Purchasing oncology services. Cancer 88: 2876-2886, 2000[CrossRef][Medline] 3. Hewitt M , Simone JV (eds): Ensuring Quality Cancer Care. Washington DC, National Cancer Policy Board, Institute of Medicine and National Research Council, National Academy Press, 1999 4. Lohr KN: Medicare: A Strategy for Quality Assurance—Committee to Design a Strategy for Quality Review and Assurance in Medicare. Washington DC, Institute of Medicine, Division of Health Care Services, National Academy Press, 1990 5. Donabedian A: Promoting quality through evaluating the process of patient care. Med Care 6: 181-202, 1968[CrossRef] 6. Donabedian A: The Definition of Quality and Approaches to Its Assessment. Ann Arbor MI, Health Administration Press, 1980
7. Potosky AL, Merrill RM, Riley GF, et al: Breast cancer survival and treatment in health maintenance organization and fee-for-service settings. J Natl Cancer Inst 89: 1683-1691, 1997 8. Farrow DC, Hunt WC, Samet JM: Geographic variation in the treatment of localized breast cancer. N Engl J Med 326: 1097-1101, 1992[Abstract] 9. Young WW, Marks SM, Kohler SA, et al: Dissemination of clinical results: Mastectomy versus lumpectomy and radiation therapy. Med Care 34: 1003-1017, 1996[CrossRef][Medline]
10. Lazovich D, White E, Thomas DB, et al: Underutilization of breast-conserving surgery and radiation therapy among women with stage I or II breast cancer. JAMA 266: 3433-3438, 1991
11. Ballard-Barbash R, Potosky AL, Harlan LC, et al: Factors associated with surgical and radiation therapy for early stage breast cancer in older women. J Natl Cancer Inst 88: 716-726, 1996 12. Hillner BE, McDonal KM, Penberthy L, et al: Measuring standards of care for early breast cancer in an insured population. J Clin Oncol 15: 1401-1408, 1997[Abstract] 13. Hillner BE, Penberthy L, Desch CE, et al: Variation in staging and treatment of local and regional breast cancer in the elderly. Breast Cancer Res Treat 40: 75-86, 1996[CrossRef][Medline] 14. Guadagnoli E, Shapiro CL, Weeks JC, et al: The quality of care for treatment of early stage breast carcinoma: Is it consistent with national guidelines? Cancer 83: 302-309, 1998[CrossRef][Medline] 15. Guadagnoli E, Shapiro C, Gurwitz JH, et al: Age-related patterns of care: Evidence against ageism in the treatment of early-stage breast cancer. J Clin Oncol 6: 2338-2344, 1997 16. Jessup JM, Menck HR, Winchester DP, et al: The National Cancer Data Base Report on patterns of hospital reporting. Cancer 78: 1829-18237, 1996[CrossRef][Medline] 17. Ries LAG , Eisner MP, Kosary CL, et al (eds): SEER Cancer Statistics Review, 1973-1997. Bethesda MD, National Cancer Institute, 2000
18. Potosky AL, Merrill RM, Riley GF, et al: Breast cancer survival and treatment in health maintenance organization and fee-for-service settings. J Natl Cancer Inst 89: 1683-1691, 1997
19. Potosky AL, Warren JL: Radical prostatectomy: Does higher volume lead to better quality? J Natl Cancer Inst 91: 1906-1907, 1999
20. Yao S-L, Yao G: Population-based study of relationships between hospital volume of prostatectomies, patient outcomes, and length of hospital stay. J Natl Cancer Inst 91: 1950-1956, 1999 21. Heckman J: Sample selection bias as a specification error. Econometrica 47: 153-161, 1979[CrossRef] 22. Cooper GS, Zhong Y, Stange KC, et al: The sensitivity of Medicare claims data for case ascertainment of six common cancers. Med Care 37: 436-444, 1999[CrossRef][Medline]
23. McClish DK, Penberthy L, Whittemore M, et al: Ability of Medicare claims data and cancer registries to identify cancer cases and treatment. Am J Epidemiol 145: 227-233, 1997 24. Warren JL, Feuer E, Potosky AL, et al: Use of Medicare hospital and physician data to assess breast cancer incidence. Med Care 37: 445-456, 1999[CrossRef][Medline] 25. Warren JL, Riley GAF, McBean AM, et al: Use of Medicare data to identify incident breast cancer cases. Health Care Financ Rev 18: 237-246, 1996[Medline] 26. Freeman JL, Zhang D, Freeman DH, et al: An approach to identifying incident breast cancer cases using Medicare claims data. J Clin Epidemiol 53: 605-614, 2000[CrossRef][Medline]
27. McClish DK, Penberthy L, Whittemore M, et al: Ability of Medicare claims data and cancer registries to identify cancer cases and treatment. Am J Epidemiol 145: 227-233, 1997 28. Solin LJ, MacPherson S, Schultz DJ, et al: Evaluation of an algorithm to identify women with carcinoma of the breast. J Med Syst 21: 189-199, 1997[CrossRef][Medline] 29. Leung K, Hasan AG, Rees KS, et al: Patients with newly diagnosed carcinoma of the breast: Validation of a claim-based identification algorithm. J Clin Epidemiol 52: 57-65, 1999[CrossRef][Medline]
30. Bickell NA, Chassin MR: Determining the quality of breast cancer care: Do tumor registries measure up? Ann Intern Med 132: 705-710, 2000 31. Institute of Medicine, Division of Health Care Services: Committee on the Role of Institutional Review Boards in Health Services Research Data Privacy Protection: Protecting Data Privacy in Health Services Research. Washington DC, National Academy Press, 2000
32. Casarett D, Karlawish JHT, Sugarman J: Determining when quality improvement initiatives should be considered research. JAMA 283: 2275-2280, 2000 33. National Program of Cancer Registries: Cancer Surveillance System (NPCR-CSS) Rationale and Approach. Atlanta GA, National Program of Cancer Registries, 1999 34. Young JL: The hospital-based cancer registry, in Jensen OM, Parkin DM, Maclennan R, et al (eds): Cancer Registration Principles and Methods. Lyon France, International Agency for Research on Cancer, 1991, pp 177-184 35. Jensen OM, Storm HH: Purposes and uses of cancer registration, in Jensen OM, Parkin DM, Maclennan R, et al (eds): Cancer Registration Principles and Methods. Lyon France, International Agency for Research on Cancer, 1991, pp 7-21 36. Austin DF: Types of registries: Goals and objectives, in Menck H, Smart C (eds): Central Cancer Registries: Design, Management, and Use. Reading United Kingdom, Hardwood Academic Publishers, 1994, pp 1-12 37. Gordon B: Data set planning, in Menck H, Smart C (eds): Central Cancer Registries: Design, Management, and Use. Reading United Kingdom, Hardwood Academic Publishers, 1994
38. Warren JL, Riley GF, Potosky AL, et al: Trends and outcomes of outpatient mastectomy in elderly women. J Natl Cancer Inst 90: 833-840, 1998 39. American Hospital Association Guide to the Health Care Field, 1997-98 edition. Chicago IL, American Hospital Association, 1997 40. Partridge EE: The National Cancer Database: Ten years of growth and commitment. CA Cancer J Clin 48: 131-133, 1998[Abstract] 41. Menck HR, Bland KI, Eyre HJ, et al: Clinical highlights from the national cancer data base. CA Cancer J Clin 48: 134-136, 1998[Abstract] 42. Zippin C, Lum D, Hankey BF: Completeness of hospital cancer case reporting from the SEER program of the National Cancer Institute. Cancer 76: 2343-2350, 1995[CrossRef][Medline] 43. National Cancer Institute: About SEER. Available at: http://www.seer.ims.nci.nih.gov/AboutSEER.html. Accessed April 2001 44. Miller M, Swan J: SEER doubles coverage by adding registries for four states [Oxford Journals Online]. J Natl Cancer Inst 93:500 [2 screens]. Available at: http://www.seer.ims.nci.nih.gov. Accessed April 4, 2001 45. State Cancer Registries: Status of Authorizing Legislation and Enabling Regulations—United States, October, 1993. MMWR Morb Mortal Wkly Rep 43:73, 74-75, 1994 46. Centers for Disease Control: At-A-Glance: The National Program of Cancer Registries. Atlanta GA, Centers for Disease Control and Prevention, 2001 47. Hewitt M, Simone J: Enhancing Data Systems to Improve the Quality of Cancer Care. Washington DC, National Cancer Policy Board, Institute of Medicine and National Research Council, 2000 48. Hillner B, Mdonal K, Penberthy L, et al: Measuring standards of care for early breast cancer in an insured population. J Clin Oncol 15: 1401-1408, 1997[Abstract] 49. National Program of Cancer Registries: Cancer Surveillance System (NPCR-CSS) Rationale and Approach. Atlanta GA, National Program of Cancer Registries, 1999 50. North American Association of Cancer Registries. North American Association of Cancer Registries Web site [Online Source]. Available at: http://www.naar.org/Certification/1998-Registries.html. Accessed August 10, 2001 51. Ma M, Bell J, Campbell S, et al: Breast cancer management: Is volume related to quality? Br J Cancer 75: 1652-1659, 1997[Medline] 52. The American Society of Clinical Oncology: Questions and Answers for ASCO’s National Initiative on Cancer Care Quality (NICCQ). Available at: http://www.asco.org/prof/pp/html/m_niccq-qa.htm. Accessed July 3, 2002 53. University of Southern California School of Medicine: Cancer Surveillance Program Procedure Guide. Los Angeles CA, University of Southern California School of Medicine, 1996 54. Minnesota Department of Health: Adopted Permanent Rules Relating to Cancer Surveillance system. Minnesota Rules: Chapter 4606: August 7, 1997 55. Aldrich TE, Vann D, Moorman PG, et al: Rapid reporting of cancer incidence in a population-based study of breast cancer: One constructive use of a central cancer registry. Breast Cancer Res Treat 35: 61-64, 1995[CrossRef][Medline] 56. Cancer Care Outcomes Research and Surveillance Consortium. Available at: http://www.healthservices.cancer.gov/cancors.html. Accessed October 17, 2001
57. Potosky AL, Harlan LC, Stanford JL, et al: Prostate cancer practice patterns and quality of life: The Prostate Cancer Outcomes Study. J Natl Cancer Inst 91: 1719-1724, 1999 58. Silliman RA, Troyan SL, Guadagnoli E, et al: The impact of age, marital status, and physician-patient interactions on the care of older women with breast carcinoma. Cancer 80: 1326-1334, 1997[CrossRef][Medline]
59. Fletcher SW, Black W, Harris R, et al: Report of the international workshop on screening for breast cancer. J Natl Cancer Inst 85: 1644–1656, 1993 60. Sackett DL: Bias in analytic research. J Chronic Dis 32: 51-63, 1979[CrossRef][Medline] 61. Hillner BE, MacDonald MK, Penberthy L, et al: Measuring standards of care for early breast cancer in an insured population. J Clin Oncol 15: 1401-1408, 1997[Abstract] 62. Abelson RP, Loftus EF, Greenwald AG: Attempts to improve the accuracy of self-reports of voting, in Tanus JM (ed): Questions About Questions: Inquiries Into the Cognitive Bases and Surveys. New York NY, Russell Sage Foundation, 1991 63. Kahn KL, Malin JL, Adams J, et al: Developing a reliable, valid, and feasible plan for quality of care measurement for cancer: What should we measure? Med Care 40: III73-III85, 2002[Medline] 64. Cancer Care Outcomes Research and Surveillance Consortium (CanCORS) Online Source. Available at: http://www.cancors.org/Welcome.html. Accessed October 15, 2001
65. Goldman DP, Schoenbaum ML, Potosky AL, et al: Measuring the incremental cost of clinical cancer research. J Clin Oncol 19: 105-110, 2001
66. Wagner JL, Alberts SR, Sloan JA, et al: Incremental costs of enrolling cancer patients in clinical trials: A population-based study. J Natl Cancer Inst 91: 847-853, 1999 Submitted May 8, 2001; accepted July 22, 2002.
This article has been cited by other articles:
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|
|||||||||||
|
|||||||||||
|
|
Copyright © 2002 by the American Society of Clinical Oncology, Online ISSN: 1527-7755. Print ISSN: 0732-183X
|
TOP
ABSTRACT
INTRODUCTION
