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 131
Enhancing DATA SYSTEMS to Improve the Quality of CANCER Care APPENDIX C Summary of Selected Registry-Based Quality Studies
OCR for page 132
Enhancing DATA SYSTEMS to Improve the Quality of CANCER Care TABLE C-1 Examples of Cancer Care Quality Studies Using State Cancer Registry Data (published in the last 10 years) Author State/Topic Design Conclusion Comments Wanebo HJ, Cole B, Chung M, et al. Is Surgical Management Compromised in Elderly Patients with Breast Cancer? Annals of Surgery 225(5):579–589, 1997. Rhode Island Patterns of care associated with age among women treated surgically for breast cancer Screening and treatment patterns assessed for 5,962 women diagnosed with breast cancer between 1987 and 1995 identified through the state registry (study limited to the 9 institutions using AJCC tumor classification). Descriptive statistics for treatment by age and stage (no comorbidity measures). Breast cancer management is compromised in the elderly. Detection rate of preinvasive cancers in women 65+ was 8.8% vs. 13.7% for women 40–65. Lumpectomy alone was done in 25.0% of elderly patients with stage I cancer vs. 9.5% in patients 40–65. Lumpectomy alone was done in 9.5% of stage II and 10.6% of stage III in patients 65+ vs. 2.7% and 2.2%, respectively, in younger patients. The study was not population-based and was limited to those institutions with AJCC tumor classification. No data on comorbidity. available. Adams-Cameron M, Gilliland FD, Hunt WC, et al. Trends in Incidence and Treatment for Ductal Carcinoma in Situ in Hispanic, American Indian, and Non-Hispanic White Women in New Mexico, 1973–1994. Cancer 85(5):1084–1090, 1999. New Mexico Patterns of care associated with race/ ethnicity among women with ductal carcinoma in situ (DCIS) Treatment patterns assessed for 950 cases of DCIS identified through the state cancer registry from 1973 to 1994. Patient characteristics included: age at diagnosis, ethnicity, residence, and poverty status (as determined by census tract of residence). Physician characteristics included: age, gender, specialty, volume of surgical breast carcinoma patients, and location of treatment. The use of BCS for DCIS increased to 52% by 1994. Geographic location of treatment was the most significant predictor of treatment. Other patient and provider characteristics were not related to use of breast-conserving surgery (BCS). The variation in rates of BCS by treatment location most likely reflects differences in physician practices and treatment recommendations.
OCR for page 133
Enhancing DATA SYSTEMS to Improve the Quality of CANCER Care Hand R, Sener S, Imperato J, et al. Hospital Variables Associated with Quality of Care for Breast Cancer Patients. JAMA 226(24):3429–3432, 1991. Illinois Hospital characteristics potentially affecting compliance with clinical standards for care of breast cancer Analysis of cancer registry data on 5,766 patients diagnosed in 1988, treated at 99 Illinois hospitals. Five quality indicators studied: proportion of patients diagnosed at late stage, hormone receptor determination, adjuvant therapy, radiation therapy, and axillary lymph node dissection. Urban location, small size, and marginal reimbursement were related to late diagnosis and inadequate treatment of breast cancer at some hospitals. Nonsignicifant hospital variables were proportion of oncology cases and teaching status of hospital. Few teaching hospitals were included. The study did not address individual patient or physician factors, or lack of payment by insurance. Concurrence on efficiency of treatments and delayed incorporation of new findings into standard care may have influenced hospital practices. Some hospital factors, such as size, setting, and reimbursement, may limit the resources available for cancer care. Desch CE, Penberthy L, Newschaffer CJ, et al. Factors that Determine the Treatment for Local and Regional Prostate Cancer. Medical Care 34 (2):152–162, 1996. Virginia Significance of comorbid and nonclinical factors in prostate cancer care Three treatment alternatives evaluated for 3,117 men diagnosed between 1985 and 1989: treatment vs. no treatment, surgery vs. radiation, hormonal/orchiectomy vs. surgery/radiation. In addition to comorbidity, other nonclinical factors studied were age, race, residence, and socioeconomic status. Data from linkages of state registry to Medicare claim files, the Area Resource File, and 1990 Census Data. Age was the most important factor in treatment decisions, even after adjustments for comorbidity. Older men received less treatment as compared with any treatment, less surgery than radiation, and more hormonal therapy than surgery and/or radiation. Other nonclinical factors did impact treatment choices, but to a lesser degree. Limitations of the data include imprecise staging variables, lack of data on Virginia residents treated in bordering states, and lack of information on physician attributes or patient preferences.
OCR for page 134
Enhancing DATA SYSTEMS to Improve the Quality of CANCER Care Polednak AP. Predictors of Breast-Conserving Surgery in Connecticut, 1990–1992. Annals of Surgical Oncology 4(3):259–263, 1997. Connecticut Sociodemographic characteristics associated with use of BCS Analysis of predictors of BCS use (poverty status, age at diagnosis, race, marital status, extent of disease, year of diagnosis, and town of residence) among 5,266 women diagnosed in 1990–1992 with early-stage breast cancer. Data from state registry and census tract. BCS was not associated with poverty level of area of residence, but was lower for larger or node-positive cancers. Attitudes and practices of local physicians were hypothesized as being important in explaining variation in BCS use by town of residence. High BCS rates (69–94% vs. 49% statewide) were found for residents of a cluster of seven contiguous towns associated with a single hospital. Polednak AP, Shevchenko IP, Flannery JT, et al. 1996. Estimating Breast Cancer Treatment Charges in Connecticut. Connecticut Medicine 60(5): 263–267, 1996. Connecticut Treatment costs for breast cancer Study assessed charges, which generally exceed actual costs or payments. Random sample of 407 breast cancer patients (all ages) diagnosed in 1991 identified in cancer registry and linked to hospitaldischarge database through 1993; 93% of cases linked. Average charges declined with age, increased with extent of disease (stage at diagnosis), and increased with extent of surgery. The linked database is most useful in estimating charges for cancers treated mainly by surgery (including ambulatory surgery at hospitals) and for charges associated with comorbid conditions and terminal care. Radiotherapy and most chemotherapy are provided on an outpatient basis, and charges were not reflected in the hospital discharge database.
OCR for page 135
Enhancing DATA SYSTEMS to Improve the Quality of CANCER Care Hillner BE, McDonald MK, Penberthy L, et al.. Measuring Standards of Care for Early Breast Cancer in an Insured Population. Journal of Clinical Oncology 15(4):1401–1408, 1997. Virginia Process of care for women with early breast cancer 918 women with local/regional invasive breast cancer identified through linking the state registry, 1989–1991, and procedural and hospital claims from Blue Cross Blue Shield (BCBS) of Virginia. Standards of care based upon consensus conferences and literature reviews, quality targets established by the authors. Achievment of objectives or performance standards varied. For women age 50 or younger, 85% with positive axillary nodes had chemotherapy claims. For older women with positive axillary nodes, 53% had chemotherapy claims. 79% of women had a follow-up mammography within the first 18 months postoperatively. The state registry collected data on a voluntary basis from approximately 50 hospitals representing about 85% of the state's hospital beds. Some limitations of claims data for quality assessment include: lost claims, the potential bundling of services in the hospital claim, and changes in individual's insurance coverage. Smith TJ, Penberthy L, Desch CE, et al. Differences in Initial Treatment Patterns and Outcomes of Lung Cancer in the Elderly. Lung Cancer 13:235–252, 1995. Virginia Patterns of care mong elderly persons with lung cancer Incident cases of non-small-cell lung cancer (NSCLC) from the state cancer registry, 1985–1989, were matched with claims from Medicare Part a and B (n = 4,999), census tract data, and the Area Resource File. Multiple logistic regression analyses used to identify factors associated with therapy choices (controlling for comorbidity). Older age increased the likelihood of exclusion from potentially curative surgery, even after controlling for other factors. Less than 10% of patients had TNM staging; however, the locoregionaldistant classification works well to estimate survival for lung cancer.
OCR for page 136
Enhancing DATA SYSTEMS to Improve the Quality of CANCER Care Hillner B, Penberthy L, Desch CE, et al. Variation in Staging and Treatment of Local and Regional Breast Cancer in the Elderly. Breast Cancer Research and Treatment 40:75–86, 1996. Virginia Patterns of care among elderly women with local and regional breast cancer State cancer registry data were linked with Medicare claims and 1990 census data. 3,361 women identified with pathologic confirmed local and regional breast cancer in 1985–1989. Processes of care assessed: tumor size determination, axillary lymph node dissection, use of adjuvant therapy, radiation if BCS was performed. Multivariate analyses including comorbidity. Older women were more likely to present with larger tumors but were less likely to undergo an axillary node dissection, to receive chemotherapy, or to receive radiation therapy if treated with BCS. The state registry collected data on a voluntary basis from approximately 50 hospitals representing about 85% of the state's hospital beds. Only 55% of cases had complete TNM staging. A summary staging system (local, regional, and distant disease) was used from the registry. Polednak AP and Flannery JT. Black Versus White Racial Differences in Clinical Stage at Diagnosis and Treatment of Prostatic Cancer in Connecticut. Cancer 70(8):2152–2158, 1992. Connecticut Racial differences in patterns of prostate cancer care First course of treatment examined by clinical stage for black (localized, n = 133; metastases, n = 102) and white (localized n = 2,653; metastases, n = 1,083) men diagnosed with prostate cancer from 1985–1988. Data from CT cancer registry. There is little difference in therapy received by black versus white patients with prostatic cancer at a given stage of diagnosis. SEER registries limited by inadequate information on clinical stage and comorbidity. This reduces the ability to interpret the appropriateness of treatment (or lack of treatment) in comparison with ad hoc studies using hospital records.
OCR for page 137
Enhancing DATA SYSTEMS to Improve the Quality of CANCER Care Katz SJ, Hislop G, Thomas DB, et al. Delay from Symptom to Diagnosis and Treatment of Breast Cancer in Washington State and British Columbia. Medical Care 1(3):264–268, 1993. Washington, British Columbia (BC), Canada Delay in time between symptom to diagnosis for women with breast cancer 10% random sample of women 35–80 years old diagnosed with invasive nonmetastatic breast cancer in 1988 in western WA (n = 174) and BC (n = 195). Data from population-based registries assessed for: (1) time from first appearance of symptoms (date of physical exam (PE) or mammography for those without symptoms) to first physician contact; (2) time from first contact to definitive diagnosis by either aspiration or biopsy; and (3) time from definitive diagnosis to initial surgery. Information from medical record review. Median delay times from first symptom to definitive treatment were short and similar in the two regions. 13.4% of women in Washington and 4.6% of women in BC experienced a diagnosis delay of three months or longer. The higher rate in Washington is explained by greater use of screening mammography and a higher frequency of nonsuspicious diagnostic mammograms. Because this study only included women who were ultimately diagnosed with breast cancer, the results cannot be generalized to all women presenting with breast-related problems. Another limitation to the study is that medical records may not accurately reflect the onset of symptoms. The impact of delay on prognosis remains uncertain, but delays have negative psychological consequences for women during the diagnostic process. Many patients, for example, have anxiety provoked by equivocal mammographic findings. Physicians are at risk for malpractice claims as a result of a delayed diagnosis of breast cancer.
OCR for page 138
Enhancing DATA SYSTEMS to Improve the Quality of CANCER Care Howe HL, Katterhagen JG, Yates J, et al. Urban–rural Differences in the Management of Breast Cancer. Cancer Causes and Control 3:533–539, 1992. Illinois Urban–rural differences in patterns of breast cancer care Cancer management described for women diagnosed with breast cancer in 1986–1987, relative to standards described in NCI's Physician Data Query. Cases grouped as follows: 147 rural residents treated in a local hospital, 119 rural residents treated in an urban hospital, and 451 urban residents treated in a local hospital. Cases identified through the state registry, and management data were obtained via hospital record audit and physician survey. Odds ratios for treatment use adjusted for age and stage. Rural cases diagnosed in rural hospitals were less likely than urban patients to have staged tumors and more likely to have node dissections. Rural cases traveling to urban centers were less likely to have limited surgery, hormone therapy, and a biopsy as a first step surgical procedure, and more likely to have node dissection. Differential urban–rural access to state-of-the-art care contributes to the differential urban–rural rates in breast-cancer case fatality. Referral networks and potential clustering of specific management practices were assessed by individual physician, surgeon, and hospital of diagnosis. Six of the 61 rural physicians saw one third of the rural patients. Their management practices were similar to other rural physicians except for axillary node dissection and hormone therapy.
OCR for page 139
Enhancing DATA SYSTEMS to Improve the Quality of CANCER Care Ayanian JZ, Kohler BA, Abe T, et al. The Relation Between Health Insurance Coverage and Clinical Outcomes among Women with Breast Cancer. New England Journal of Medicine 329(5):326–331, 1993. New Jersey Effect of insurance status on clinical outcomes for women with breast cancer Stage of disease and stagespecific survival assessed among 4,675 women, 35–64 years of age, diagnosed with invasive breast cancer between 1985 and 1987. Data from linking state registry records to hospital-discharge data. Survival assessed through 1992. Uninsured patients and those covered by Medicaid presented with more advanced disease than did privately insured patients. Survival was worse for uninsured patients and those with Medicaid coverage than for privately insured patients with local disease and regional disease, but not distant metastases. Treatment patterns not assessed. Mann BA, Samet JM, Hunt WC et al. Changing Treatment of Breast Cancer in New Mexico From 1969 Through 1985. JAMA 259(23):3413–3417, 1988. New Mexico Temporal changes in patterns of breast cancer care Treatment within four months of diagnosis was assessed for 6,030 women diagnosed with primary breast cancer (local or regional) from 1969 through 1985 as identified in the state cancer registry. To assure accurate classification of the use of radiotherapy, files of patients recorded as having undergone BCS or simple mastectomy without radiotherapy were reviewed. In addition, records of radiotherapy facilities within the state were reviewed. No information available on comorbidity. Use of BCS rose from 6% to 25% after 1980. Women younger than 50 or older than 80 were most likely to undergo BCS. Radiotherapy after BSC could not be documented for 26% of women 65 years old or younger, or for 56% of the women aged 65 years or older. The behavior of individual physicians was assessed. During 1981 to 1984, all but 7 of 43 surgeons actively involved in breast cancer surgery had performed at least one BCS. Twenty surgeons used the procedure in >10% of their patient population. The use of BCS did not correlate with the surgeon's age, volume of cases, or use of BCS before 1981.
OCR for page 140
Enhancing DATA SYSTEMS to Improve the Quality of CANCER Care Roohan PJ, Bickell NA, Baptiste MS, et al. Hospital Volume Difference and Five-Year Survival from Breast Cancer. American Journal of Public Health 88(3):454–457, 1998. New York Effect of hospital volume of BCS cases on the 5-year survival of women treated for breast cancer Assessment of 5-year survival and risk of death for 47,890 women, diagnosed between 1984 and 1989, identified through linkages between the NY hospital discharge database and the state registry. Adjustments made for surgery type, cancer stage, comorbidity, age, race, socioeconomic status, and distance to hospital. Hospital volume classified in four groups, ranging from very low (10 or fewer cases) to high (150 BCS performed per year). Unadjusted 5-year survival rates were significantly higher at high-volume hospitals, for each cancer stage. Patients at very low-volume hospitals had a 60% higher risk of death than patients at high-volume hospitals. Patients at low- (11–50 cases) and moderate- (51–150) volume hospitals had 30% and 19%, respectively, higher risks of dying. The “dose-response” relationship between volume and survival supports a causal relationship. Limitations of administrative databases are reflected' in limitations of the study. Socioeconomic status was not measured as the individual level. It was based on contextual data—address of residence. Comorbidity measurements may not accurately reflect severity, so adjustment may have been incomplete.
Representative terms from entire chapter: