MEDICARE S HEALTH MAINTEnance

Transcription

1 ORIGINAL CONTRIBUTION Stage at Diagnosis and Treatment Patterns Among Older Women With Breast Cancer An HMO and Fee-for-Service Comparison Gerald F. Riley, MSPH Arnold L. Potosky, PhD Carrie N. Klabunde, PhD Joan L. Warren, PhD Rachel Ballard-Barbash, MD See also p 755 and Patient Page. Context Few studies have compared patterns of care in health maintenance organization (HMO) and fee-for-service (FFS) settings. Objective To examine breast cancer stage at diagnosis and, for those at an early stage, treatment patterns for elderly women in HMO and FFS settings. Design Cancer registry data from the Surveillance, Epidemiology, and End Results (SEER) program linked to Medicare enrollment records. Settings and Participants Women aged 65 years or older residing in 11 geographic areas who were newly diagnosed as having breast cancer between 1988 and Main Outcome Measures Standardized percentage of cases diagnosed at late stages for HMO vs FFS; standardized percentage of early-stage cases undergoing initial treatment with breast-conserving surgery (BCS); and, among BCS cases, standardized percentage receiving adjuvant radiation therapy. Standardization was achieved through logistic regression, controlling for patient demographics, cancer history, county of residence, year of diagnosis, and educational attainment at the census tract level. Analyses of treatment patterns were controlled for stage at diagnosis and tumor size. Results The HMO enrollees were less likely to have breast cancer diagnosed at late stages than FFS patients (HMO, 7.6%; FFS, 10.8%; difference, 3.2% [95% confidence interval (CI), 4.2% to 2.2%]). Among early-stage cases, the percentages undergoing BCS were similar in HMO and FFS settings overall (HMO, 38.4%; FFS, 36.8%; difference, 1.6% [95% CI, 0.0%-3.2%]); percentages varied markedly at the individual plan level. Among women undergoing BCS, HMO enrollees were significantly more likely to receive radiation therapy but, again, results varied by plan (HMO, 69.0%; FFS, 63.7%; difference, 5.3% [95% CI, 2.9%-7.7%]). In general, use of BCS and radiation therapy was substantially higher than that found in an earlier study examining cases diagnosed between 1985 and Conclusions Treatment of early-stage breast cancer in HMOs often differs from local FFS patterns, but not in a consistent way. During the period of our study, elderly HMO enrollees did not appear to have systematic access problems with adjuvant radiation therapy following BCS compared with women in an FFS setting. JAMA. 1999;281: MEDICARE S HEALTH MAINTEnance organization (HMO) program has grown rapidly in recent years, with more than 6.5 million persons currently enrolled, accounting for 17% of the beneficiary population. Some policymakers have promoted HMOs as a cost containment strategy, given prior research showing that HMOs use health care resources, particularly inpatient hospital services, less intensively. 1-4 Other policymakers have concerns about quality, access, and satisfaction with care among HMO enrollees; studies have yielded mixed results on the process and outcomes of HMO care Such concerns are expected to grow as the Balanced Budget Act of 1997 increases the kinds of capitated plans available under Medicare. Knowledge about how patterns of care differ between HMO and fee-for-service (FFS) settings is limited. Higher use of breast-conserving surgery (BCS) and radiation therapy has been documented among elderly women with early-stage breast cancer in 2 HMOs compared with local FFS settings. 7 The use of BCS has been shown to be slightly lower among the nonelderly in HMOs. 14 Other studies have reported treatment differences between HMO and FFS settings, with patients in HMOs less likely to undergo surgery and other invasive procedures for Author Affiliations: Health Care Financing Administration, Baltimore, Md (Mr Riley); and the National Cancer Institute, Bethesda, Md (Drs Potosky, Klabunde, Warren, and Ballard-Barbash). conditions other than breast cancer More consistent evidence is available demonstrating that HMO enrollees receive more preventive services, including can- Corresponding Author and Reprints: Gerald F. Riley, MSPH, Health Care Financing Administration, 7500 Security Blvd, Room C , Baltimore, MD ( griley@hcfa.gov). 720 JAMA, February 24, 1999 Vol 281, No American Medical Association. All rights reserved.

2 cer screening services, than persons in the FFS setting. 4,19,20 Breast cancer treatment patterns in HMO and FFS settings are of particular interest because of the widespread use of both mastectomy and BCS (with or without adjuvant radiation therapy) as primary treatments. Randomized clinical trials have demonstrated that BCS followed by breast irradiation has equivalent survival benefits to mastectomy in the treatment of early-stage breast cancer A 1990 National Institutes of Health consensus development conference concluded that breast conservation treatment is usually preferable to mastectomy because it preserves the breast. 24 The consensus conference also concluded that among patients undergoing breast conservation treatment, no subgroups have been identified in which radiation therapy can be avoided. Despite these conclusions, disagreement remains about whether very elderly women should always receive radiation therapy following BCS. 25,26 We examined treatment patterns for elderly women with early-stage breast cancer, comparing women enrolled in HMOs with those in an FFS setting. We also compared breast cancer stage at diagnosis for elderly women in HMO and FFS settings. A previous study showed that elderly women in HMOs tended to be diagnosed as having breast cancer at earlier stages than women in FFS settings between 1985 and During that time, screening mammograms were not covered under Medicare FFS but were offered by most HMOs to their Medicare enrollees as an extra benefit. We present updated findings on stage of breast cancer at diagnosis based on HMO and FFS cases diagnosed in 1991 through 1993, which capture the period after January 1991, when Medicare began covering biennial screening mammograms. METHODS Data We used cancer registry data from the National Cancer Institute s Surveillance, Epidemiology, and End Results (SEER) program linked to Medicare enrollment records. The SEER program receives uniformly reported data from 11 population-based cancer registries covering 14% of the US population. 28 For each reported case, SEER receives information on month and year of diagnosis, cancer site, patient demographics, extent of disease at diagnosis, and therapy administered within 4 months of diagnosis. Each SEER case is assigned a stage at diagnosis by the National Cancer Institute based on extent of disease (tumor size, nodal involvement, and metastasis). The SEER program includes cases diagnosed from 1973 onward, with the exception of the Los Angeles and San Jose, Calif, registries, which joined SEER recently and had reported cases only for 1992 and 1993 at the time of our study. The SEER data through 1993 have been linked to Medicare administrative records as well as census data at the census tract and ZIP code levels. 29 Sample Selection We identified women newly diagnosed as having breast cancer at age 65 years or older between 1988 and 1993 who were entitled to Medicare Parts A and B at the time of diagnosis. For women having multiple breast cancer diagnoses, we selected for analysis the first breast cancer diagnosed at age 65 years or older. We excluded cases diagnosed through death certificate review or autopsy (1% of cases). Using Medicare enrollment records, we identified women who were members of an HMO at the time of diagnosis and those in the FFS setting. Women in the FFS setting who resided in counties with no HMO enrollees were excluded from the analysis to enhance the validity of the HMO-FFS comparisons. The sample is described in TABLE 1. The analyses of stage at diagnosis and treatment patterns were based on 2 different subsamples. For stage at diagnosis, we selected cases diagnosed from 1991 through 1993 to measure differences between HMO and FFS settings subsequent to Medicare coverage of screening mammograms. After excluding 1610 cases of unknown stage, cases remained (3519 HMO and FFS cases). For the analysis of BCS vs mastectomy, we used data from 1988 through 1993 and limited the sample to stage I and II cases as defined in the staging criteria of the American Joint Committee on Cancer. 30 We excluded cases of bilateral disease, those in which no mass was found, those in which the patient switched from HMO to FFS setting or vice versa within 4 months of diagnosis, and those not treated by either BCS or total/modified radical mastectomy. After excluding 1106 cases, cases remained for analysis (4011 HMO and FFS cases). Breast-conserving surgery was defined as segmental/subtotal mastectomy, lumpectomy, quadrantectomy, tylectomy, wedge resection, nipple resection, excisional biopsy, or partial mastectomy not otherwise specified. Total/modified radical mastectomy was defined as subcutaneous mastectomy, simple mastectomy, and modified radical mastectomy. For the analysis of adjuvant radiation therapy following BCS, we omitted the 3.3% of BCS cases in which receipt of radiation therapy was unknown or refused by the patient. Analysis We used logistic regression to compare HMO and FFS enrollees regarding the probability of having breast cancer diagnosed at a late stage (stages III and IV). Model covariates were age (categorized as ages 65-69, 70-74, 75-79, 80-84, 85-90, or 91 years), race (white, black, other, or unknown), state and county of residence, cancer history (breast cancer, other cancer, or no cancer), year of diagnosis, and percentage of adults with fewer than 12 years of education at the census tract level (categorized as quartiles of low to high education level). These variables have been found to be associated with stage at diagnosis and treatment patterns in earlier studies. 27,31,32 Logistic regression was also used to compare patients with early-stage disease in HMO and FFS settings regarding the use of BCS vs mastectomy. Model covariates included those described herein as well as a group of binary variables indicating stage at diagnosis (stages 1999 American Medical Association. All rights reserved. JAMA, February 24, 1999 Vol 281, No

3 I, IIA, and IIB) and, within stage, the size of the tumor. Stage and tumor size were previously found to be associated with treatment patterns. 31 A similar model was estimated to analyze the receipt of radiation therapy among early-stage BCS cases in HMO and FFS settings. The treatment pattern models are summarized in TABLE 2. We examined differences in HMO effects across registries by testing Table 1. Number and Characteristics of Elderly Women Diagnosed as Having Breast Cancer and Residing in SEER Areas, * Characteristics HMO FFS P Value Sample size, No Average age at diagnosis, y Percentage of adults with 12 y of school, Geographic area San Francisco/Oakland, Calif Los Angeles, Calif Seattle, Wash Hawaii New Mexico Iowa Connecticut San Jose, Calif Detroit, Mich Utah Atlanta, Ga Setting Urban Rural Race White Black Other Unknown Year of diagnosis Cancer history No previous cancer diagnosis Previous breast diagnosis Previous diagnosis of other cancer Previous diagnosis of cancer, site unknown Stage at diagnosis In situ I II III IV Unknown Therapy (stages I and II) Mastectomy (simple or modified radical) Breast-conserving surgery Other/unknown Radiation therapy (breast-conserving surgery cases only) Yes No *SEER indicates Surveillance, Epidemiology, and End Results program; HMO, health maintenance organization; and FFS, fee-for-service setting. All data are presented as percentage of distribution unless indicated otherwise. Data were measured at census tract level. The value was unknown for 1% of HMO and 2.6% of FFS cases. Cases are from only. Applies to cases diagnosed in only. the statistical significance of a set of HMO-by-registry interaction terms and investigating registry-specific models. We also examined treatment patterns for the 10 largest HMOs in our database and compared them with treatment patterns in the local FFS settings. Our findings are presented as standardized percentages of patients in HMO and FFS settings who underwent BCS. The percentages were standardized to the distribution of covariates for the combined sample of early-stage cases in both settings. (That is, we used the regression model to compute predicted probabilities of undergoing BCS for each person in the combined sample, with and without the effect of HMO membership. Average predicted probabilities were computed for the combined sample, with and without the HMO effect, and multiplied by 100 to yield percentages.) A similar method was used to present results of the analyses of radiation therapy and of late stage at diagnosis. RESULTS Sample Description Women in HMOs who were diagnosed as having breast cancer tended to be significantly younger and more often of nonwhite and nonblack races than women in the FFS setting (Table 1). The HMO sample was concentrated in a few registries, reflecting heavy HMO penetration in those areas. Proportionately more HMO cases were diagnosed in 1992 and 1993, reflecting a large HMO presence in Los Angeles, which reported cases to SEER only in those years. In HMOs, 6.8% of breast cancer cases were diagnosed at late stages compared with 10.2% in the FFS setting, without adjustment. Among early-stage cases, a higher percentage of HMO enrollees (41.1%) underwent BCS than women in the FFS setting (36.0%). Among BCS cases, 73.5% of HMO enrollees and 62.8% of women in the FFS setting received radiation therapy. Multivariate Results After excluding cases of unknown stage and adjusting for model covariates, the standardized percentages of late-stage cases were 7.6% among HMO enrollees 722 JAMA, February 24, 1999 Vol 281, No American Medical Association. All rights reserved.

4 and 10.8% among women in the FFS setting (P)(TABLE 3). This result was consistent across most registries. Multivariate analysis of treatment patterns revealed that older women in the sample were much more likely to undergo BCS rather than mastectomy and were much less likely to receive radiation therapy following BCS (Table 2). Women living in areas with high education levels were more likely to undergo both BCS and radiation therapy. There was a strong trend toward increased use of BCS during the period of the study. There was also a significant trend toward increased use over time of radiation therapy following BCS. Based on results of the models shown in Table 2, 38.4% of HMO enrollees underwent BCS, standardized to the HMO and FFS samples combined (Table 3). The standardized percentage of FFS cases undergoing BCS was slightly lower (36.8%; P =.06). Among early-stage cases undergoing BCS, women in HMOs were significantly more likely to receive radiation therapy than women in the FFS setting (69.0% compared with 63.7%, respectively; P)(Table 3). TherewassubstantialvariationinHMO- FFS differences by registry with respect to both BCS and radiation therapy. In Seattle, Wash, San Francisco/Oakland, Calif, and San Jose, HMO enrollees were significantly more likely to undergo BCS than women in the FFS setting. Among BCS cases in Seattle and San Francisco/ Oakland, a significantly higher percentage of HMO enrollees received radiation therapythandidwomenintheffssetting. Los Angeles HMO enrollees were significantly less likely to undergo BCS than women in the FFS setting; among BCS cases there, relatively fewer HMO enrollees received radiation therapy than women in the FFS setting, but the difference was not statistically significant. With the exception of Los Angeles, HMO-FFS differences within specific registries were attributable to the influence of 1 or 2 large HMOs. Analyses at the individual plan level revealed that rates of BCS were significantly higher in some HMOs than in the local FFS setting while in other plans, rates of BCS were signifi- Table 2. Odds Ratios for the Association of Patient Characteristics With Undergoing BCS and Receipt of Radiation Therapy Among BCS Cases in Elderly Women Diagnosed as Having Stage I and II Breast Cancer and Residing in SEER Areas, * Independent Variables BCS Among Stage I and II Cases, OR (95% CI) Radiation Therapy Among BCS Cases, OR (95% CI) Age, y (...) 1.00 (...) ( ) 0.72 ( ) ( ) 0.49 ( ) ( ) 0.19 ( ) ( ) 0.07 ( ) ( ) 0.02 ( ) Race White 1.00 (...) 1.00 (...) Black 1.19 ( ) 0.61 ( ) Other 0.75 ( ) 1.12 ( ) Unknown 1.34 ( ) 0.23 ( ) Cancer history None 1.00 (...) 1.00 (...) Breast 0.42 ( ) 0.54 ( ) Other 1.25 ( ) 0.63 ( ) Site unknown 0.64 ( ) 0.72 ( ) Education level Highest quartile 1.00 (...) 1.00 (...) Second-highest quartile 0.88 ( ) 0.97 ( ) Second-lowest quartile 0.80 ( ) 0.83 ( ) Lowest quartile 0.80 ( ) 0.77 ( ) Unknown 0.86 ( ) 1.24 ( ) Year of diagnosis (...) 1.00 (...) ( ) 0.99 ( ) ( ) 0.83 ( ) ( ) 0.77 ( ) ( ) 0.68 ( ) ( ) 0.73 ( ) Tumor size, cm Stage I ( ) 0.42 ( ) ( ) 0.86 ( ) ( ) 0.94 ( ) ( ) 0.87 ( ) Stage IIA ( ) 2.74 ( ) ( ) 1.28 ( ) ( ) 1.67 ( ) ( ) 0.73 ( ) ( ) 0.52 ( ) Stage IIB ( ) 1.00 (...) ( ) 1.00 (...) ( ) 1.00 (...) (...) 1.00 (...) Stage II, size unknown 0.68 ( ) 1.00 (...) HMO enrollee 1.08 ( ) 1.36 ( ) *HMO indicates health maintenance organization; BCS, breast-conserving surgery; SEER, Surveillance, Epidemiology, and End Results program; OR, odds ratio; CI, confidence interval; and ellipses, reference group. Stage I cases have no lymph node involvement. Stage IIA cases at 2.0 cm or less have lymph node involvement. Stage IIA cases at more than 2.0 cm are node-negative or unknown. Stage IIB cases at 5.0 cm or less are node-positive. Stage IIB cases at more than 5.0 cm are node-negative or unknown. Stage II cases of unknown size are node-positive. Table does not include results for binary variables indicating state and county of residence. 95% confidence interval does not include 1.0. Data were measured at census tract level as quartiles of low to high education. Microscopic focus or mammography diagnosis only, with no size given American Medical Association. All rights reserved. JAMA, February 24, 1999 Vol 281, No

5 cantly lower (TABLE 4). Among the 10 HMOs included in Table 4, 5 had higher rates of BCS than the local FFS setting and 5 had lower rates; each rate was standardized to the characteristics of the combined sample of the HMO s enrollees and FFS cases in its service area. Two HMOs had rates of BCS that were significantly higher than the rates in their local FFS areas, including 1 that exceeded the local FFS rate by 18.1%. On the other hand, 3 HMOs had rates of BCS that were 9% to 12% lower than local FFS rates. In 3 HMOs, BCS cases received radiation therapy at a rate 9% to 13.5% higher than women treated locally in the FFS setting. In a fourth HMO, BCS cases received radiation therapy at a rate 11.9% lower than FFS cases in its service area. It is noteworthy that the HMOs with high rates of BCS also tended to provide more radiation therapy than local providers in the FFS setting. The plans in Table 4 had sample sizes of between 94 and 1021 cases for analyses of treatment patterns, accounting for 77% of HMO enrollees in the entire sample. The fact that most of the HMO cases were enrolled in only 10 plans precluded an examination of treatment patterns by plan characteristics. Table 3. Standardized Percentages of Cases Diagnosed at Late Stages, Stage I and II Cases Undergoing BCS, and BCS Cases Receiving Radiation Therapy Among Elderly Women Diagnosed as Having Breast Cancer and Residing in SEER Areas, * Standardized % HMO FFS Difference 95% CI for Difference Late-stage cases to 2.2 Early-stage cases undergoing BCS to 3.2 BCS cases receiving radiation therapy to 7.7 *BCS indicates breast-conserving surgery; SEER, Surveillance, Epidemiology, and End Results program; HMO, health maintenance organization; FFS, fee-for-service setting; and CI, confidence interval. Percentages are adjusted for age, race, cancer history, year of diagnosis, county of residence, and education at the census tract level. Percentages related to treatment patterns have also been adjusted for stage at diagnosis and tumor size. Percentages have been standardized to the characteristics of the combined samples of HMO and FFS cases. Data apply to cases diagnosed in % confidence interval does not include 0.0. Table 4. Standardized Percentages of Stage I and II Cases Undergoing BCS and BCS Cases Receiving Radiation Therapy Among Elderly Women in Large HMOs Diagnosed as Having Breast Cancer and Residing in SEER Areas, * Cases Undergoing BCS, % BCS Cases Receiving Radiation Therapy, % Specific HMOs HMO FFS Difference 95% CI for Difference HMO FFS Difference 95% CI for Difference to to to to to to to to to to to to to to to to to to 16.1 *BCS indicates breast-conserving surgery; HMO, health maintenance organization; SEER, Surveillance, Epidemiology, and End Results program; FFS, fee-for-service setting; CI, confidence interval; and ellipses, data not analyzed because of insufficient sample size ( 50 HMO enrollees). Percentages are adjusted for age, race, cancer history, year of diagnosis, stage at diagnosis, tumor size, county of residence, and education at the census tract level and have been standardized to the characteristics of the combined sample of the HMO s enrollees and FFS cases in its service area. 95% confidence interval does not include 0.0. COMMENT Trends The use of BCS among elderly women with early-stage breast cancer was increasing rapidly at the time of our study. Ballard-Barbash et al 31 found that only about 13% of early-stage cases underwent BCS plus radiation therapy in SEER areas between 1985 and 1989, suggesting that BCS was not a common curative procedure for older women. Based on data from 1988 through 1993, we found that approximately 23% of early-stage cases underwent BCS plus radiation therapy. Furthermore, we found a strong secular trend for increasing use of BCS, as well as for increasing use of radiation therapy among BCS cases. These secular trends held for both HMO and FFS cases. Increased use of BCS was also observed in the early 1990s by Guadagnoli et al 33 in Massachusetts and Minnesota. Our findings suggest that BCS and radiation therapy were undergoing rapid diffusion at the time of our study and were being adopted increasingly in both the managed care and FFS sectors. Treatment Patterns We found significant differences between HMO and FFS enrollees regarding treatment patterns. Enrollees in some HMOs were significantly more likely to undergo BCS than women in the FFS setting; in other HMOs, the opposite was true. Among BCS cases, radiation therapy was more commonly received by HMO enrollees overall but, again, the pattern varied by HMO. Our findings demonstrate that aggregate comparisons of the experiences of HMO and FFS enrollees may obscure important patterns within the HMO and FFS populations. Health maintenance organizations differ in their structure, organization, benefit packages, payment policies, practice protocols, and provider relationships. Market characteristics also vary along many dimensions, such as degree of competition, managed care penetration, and availability of radiation facilities. The reasons for the treatment pattern differences we observed are likely to vary by both plan and market characteristics and may be idiosyncratic to individual HMOs. In particular, our findings illustrate the limitations of comparing treatment practices in the HMO and FFS settings, based 724 JAMA, February 24, 1999 Vol 281, No American Medical Association. All rights reserved.

6 on analyses conducted for a single geographic area or for individual plans. 6,7 The organized delivery systems embodied in some HMOs, particularly staff and group models, may facilitate dissemination of new standards of care. 34 Some HMOs are committed in principle to the practice of evidence-based medicine, ie, the provision of services that have been proven to be efficacious. Many of these plans develop practice protocols and formalized care guidelines based on published evidence. Although many FFS providers may also practice evidence-based medicine, wellmanaged HMOs with established provider networks may have greater institutional ability to encourage the dissemination and adoption of currently recommended treatments. The presence of an organized delivery system may also facilitate the provision of multiple modes of care, such as surgical treatment and adjuvant therapies. Some plans may have a culture of practice thatencouragescertainpatternsofcare that are consistent with current standards of treatment. 35 Emphasis on patient education and counseling in some plans may also play a role in the patient s decision to undergo BCS and radiation therapy. Many patients are not well informed of their treatment options for early-stage breast cancer; this may have been especially true during the time of our study, when BCS plus radiation was still an emerging therapy. To the extent that some HMOs emphasize the importance of patients understanding and choosing among treatment options, the use of BCS and radiation therapy may have increased. Treatment differences between HMO and FFS patients may also be related to selection of physicians and hospitals on the part of HMOs. For example, significant differences in propensity to use BCS have been found by type of hospital. 36 Treatment differences may also reflect preexisting knowledge, preferences, or attitudes toward seeking care on the part of beneficiaries who choose to enroll in various plans. There is concern on the part of many policymakers that financial considerations may inappropriately influence treatment patterns under capitated arrangements, particularly where provider groups are placed directly at risk. There do not appear to be strong incentives favoring use of mastectomy or BCS plus radiation therapy because the costs of these treatments are similar or, perhaps, slightly greater for the latter. 37,38 However, there is concern that radiation therapy could be underused among women undergoing BCS, given the marginal costs associated with radiation therapy. Our findings are reassuring in that they suggest that most elderly HMO enrollees are not experiencing problems with access to radiation therapy compared with women in the FFS setting. Nonetheless, we found evidence that the use of radiation therapy following BCS may be low in some HMOs. We were not able to compare patterns of use with specific financial incentives because we did not have information on the types of provider compensation used within the HMOs included in our study. Stage at Diagnosis We found that HMO enrollees continued to have breast cancer diagnosed at earlier stagesthanwomenintheffssetting,even after the initiation of Medicare coverage for screening mammograms in Severalfactorsmayexplainthispattern. Health maintenance organizations may provide more education on the availability and importance of breast cancer screening. Some plans have established screening policies for their providers 39 and some have developed centralized systems for outreach and delivery of preventive services to their enrollees. 40,41 Such outreach activitiesmay be influenced by Health Plan Employer Data and Information Set reporting requirements for HMOs, which emphasize the provision of preventive services as a measure of plan quality. In addition, many women in the Medicare FFS setting may be unaware of the screening mammography benefit. Out-of-pocket costs may also beadeterrentforwomenwithoutasupplemental health insurance policy because HMOs typically have small copayments associated with screening mammograms, whereas Medicare deductible and coinsurance amounts were in effect in the FFS setting during the years of our study. Limitations We could not control for patient comorbidity in our analyses because data on comorbidities were not available for Medicare HMO enrollees. Medicare beneficiaries in the FFS setting tend to have more diagnosed conditions and poorer health status than persons enrolled in HMOs. 42 Greater comorbidity has been shown to be associated with significantly greater use of BCS and less receipt of radiation therapy among BCS cases. 31 Our findings may be affected by completeness of documentation on extent of disease and treatment within HMO and FFS settings. An unpublished analysis (written communication, SEER program staff, 1988) estimated the reporting of radiation therapy in the SEER program to be about 90% complete, based on reabstraction of a sample of cases. We do not know whether reporting of radiation therapy and other information differs in completeness and quality between the HMO and FFS sectors. Our findings could be influenced by spillover effects between the HMO and FFS sectors, ie, the tendency for practice patterns among HMO providers to affect practice in the FFS setting and vice versa. Spillover effects would probably tend to dampen treatment pattern differences because of the tendency for care in one sector to resemble care in the other sector over time. CONCLUSION The growth of the Medicare HMO program has substantially increased the managed care options available to beneficiaries. As the number and types of plans increase with implementation of the Balanced Budget Act, it will be important to monitor the effects of alternative delivery systems on breast cancer detection and treatment. Such effects will likely continue to vary significantly by geographic area and plan. Acknowledgment: We thank Kathryn Schleeter of Information Management Services, Silver Spring, Md, for her valuable assistance in preparing data files for the project. We also thank James Lubitz, MPH, and Martin Brown, PhD, for their excellent reviews of early drafts of the manuscript American Medical Association. All rights reserved. JAMA, February 24, 1999 Vol 281, No

7 REFERENCES 1. Sturm R, Jackson CA, Meredith LS, et al. Mental health care utilization in prepaid and fee-for-service plans among depressed patients in the Medical Outcomes Study. Health Serv Res. 1995;30: Brown RS, Bergeron JW, Clement DG, Hill JW, Retchin SM. The Medicare Risk Program for HMOs: Final Summary Report on Findings From the Evaluation. Princeton, NJ: Mathematica Policy Research Inc; Final report under HCFA contract Greenfield S, Nelson EC, Zubkoff M, et al. Variations in resource utilization among medical specialties and systems of care. JAMA. 1992;267: Manning WG, Leibowitz A, Goldberg GA, Rogers WH, Newhouse JP. A controlled trial of the effect of a prepaid group practice on use of services. N Engl J Med. 1984;310: Miller RH, Luft HS. Does managed care lead to better or worse quality of care? Health Aff (Millwood). 1997;16: Lee-Feldstein A, Anton-Culver H, Feldstein PJ. Treatment differences and other prognostic factors related to breast cancer survival. JAMA. 1994;271: 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. 1997;89: Carlisle DM, Siu AL, Keeler EB, et al. HMO vs feefor-service care of older persons with acute myocardial infarction.am J Public Health. 1992;82: Clement DG, Retchin SM, Brown RS, Stegali MH. Access and outcomes of elderly patients enrolled in managed care. JAMA. 1994;271: Ware JE, Bayliss MS, Rogers WH, Kosinski M, Tarlov AR. Differences in 4-year health outcomes for elderly and poor, chronically ill patients treated in HMO and fee-for-service systems. JAMA. 1996;276: Retchin SM, Clement DG, Rossiter LF, Brown B, Brown R, Nelson L. How the elderly fare in HMOs: outcomes from the Medicare competition demonstrations. Health Serv Res. 1992;27: Safran DG, Tarlov AR, Rogers WH. Primary care performance in fee-for-service and prepaid health care systems. JAMA. 1994;271: Nelson L, Gold M, Brown R, Ciemnecki AB, Aizer A, Cybulski KA. Access to Care in Medicare Managed Care: Results From a 1996 Survey of Enrollees and Disenrollees. Washington, DC: Mathematica Policy Research Inc; Final report under Physician Payment Review Commission contract Hadley J, Mitchell JM. Breast cancer treatment choice and mastectomy length of stay: a comparison of HMO and other privately insured women. Inquiry. 1997/1998;34: Young G, Cohen B. Inequities in hospital care: the Massachusetts experience. Inquiry. 1991;28: Every NR, Fihn SD, Maynard C, Martin JS, Weaver WD. Resource utilization in treatment of acute myocardial infarction: staff-model health maintenance organization versus fee-for-service hospitals. J Am Coll Cardiol. 1995;26: Langa KM, Sussman EJ. The effect of costcontainment policies on rates of coronary revascularization in California. N Engl J Med. 1993;329: Greenwald HP, Henke CJ. HMO membership, treatment, and mortality risk among prostatic cancer patients. Am J Public Health. 1992;82: Potosky AL, Breen N, Graubard BI, Parsons PE. The association between health care coverage and the use of cancer screening tests. Med Care. 1998;36: Miller RH, Luft HS. Managed care plan performance since JAMA. 1994;271: Fisher B, Bauer M, Margolese R, et al. Five-year results of a randomized clinical trial comparing total mastectomy and segmental mastectomy with or without radiation in the treatment of breast cancer. N Engl J Med. 1985;312: Veronesi U, Saccozzi R, Del Vecchio M, et al. Comparing radical mastectomy with quadrantectomy, axillary dissection, and radiotherapy in patients with small cancers of the breast. N Engl J Med. 1981;305: Jacobson JA, Danforth DN, Cowan KH, et al. Tenyear results of a comparison of conservation with mastectomy in the treatment of stage I and II breast cancer. N Engl J Med. 1995;332: NIH Consensus Conference: treatment of earlystage breast cancer. JAMA. 1991;265: Veronesi U, Luini A, Del Vecchio M, et al. Radiotherapy after breast-preserving surgery in women with localized cancer of the breast. N Engl J Med. 1993; 328: Silliman RA, Balducci L, Goodwin JS, Holmes FF, Leventhal EA. Breast cancer care in old age: what we know, don t know, and do. J Natl Cancer Inst. 1993; 85: Riley GF, Potosky AL, Lubitz JD, Brown ML. Stage of cancer at diagnosis for Medicare HMO and feefor-service enrollees. Am J Public Health. 1994;84: Ries LAG, Kosary CL, Hankey BF, Miller BA, Harras A, Edwards BK, eds. SEER Cancer Statistics Review, , National Cancer Institute. Bethesda, Md: National Institutes of Health; NIH publication Potosky AL, Riley G, Lubitz J, Mentnech R, Kessler L. Potential for cancer related health services research using a linked Medicare-tumor registry data base. Med Care. 1993;31: American Joint Committee on Cancer. Manual for Staging of Cancer. Philadelphia, Pa: JB Lippincott; Ballard-Barbash R, Potosky AL, Harlan LC, Nayfield SG, Kessler LG. Factors associated with surgical and radiation therapy for early stage breast cancer in older women. J Natl Cancer Inst. 1996;88: Farrow DC, Hunt WC, Samet JM. Geographic variation in the treatment of localized breast cancer. N Engl J Med. 1992;326: Guadagnoli E, Weeks JC, Shapiro CL, Gurwitz JH, Barbas C, Soumerai SB. Use of breast conserving surgery for treatment of stage I and stage II breast cancer. J Clin Oncol. 1998;16: Wagner EH, Austin BT, Von Korff M. Organizing care for patients with chronic illness. Milbank Q. 1996;74: Clancy CM, Brody H. Managed care: Jekyll or Hyde? JAMA. 1995;273: Nattinger AB, Gottlieb MS, Hoffman RG, Walker AP, Goodwin JS. Minimal increase in use of breastconserving surgery from 1986 to Med Care. 1996;34: Munoz E, Shamash F, Friedman M, Teicher I, Wise L. Lumpectomy vs mastectomy. Arch Surg. 1986;121: Barlow W, Taplin S, Ichikawa L. A Cost Comparison of Treatment Alternatives for Early-Stage Breast Cancer. Seattle, Wash: Group Health Cooperative of Puget Sound; Final report under National Cancer Institute contract NO1-CN Bernstein AB, Thompson GB, Harlan LC. Differences in rates of cancer screening by usual source of medical care. Med Care. 1991;29: Trock B, Rimer BK, King E, Balshem A, Cristinzio CS, Engstrom PF. Impact of an HMO-based intervention to increase mammography utilization. Cancer Epidemiol Biomarkers Prev. 1993;2: Prevention and managed care: opportunities for managed care organizations, purchasers of health care, and public health agencies. MMWR Morb Mortal Wkly Rep. 1995;44: Riley G, Tudor C, Chiang Y, Ingber M. Health status of Medicare enrollees in HMOs and fee-forservice in Health Care Financing Rev. 1996;17: JAMA, February 24, 1999 Vol 281, No American Medical Association. All rights reserved.