review of organizational factors related to care offered to women

Epidemiologic ReviewsCopyright © 2000 by The Johns Hopkins University School of Hygiene and Public HealthAll rights reserved

Vol. 22, No. 2Printed in U.S.A.

Review of Organizational Factors Related to Care Offered to Women withBreast Cancer

Nicole Hebert-Croteau,1 Jacques Brisson,2 and Raynald Pineault3

INTRODUCTION

Recent years have seen a growing interest in quality ofcare as a means of improving medical care while controllingthe rising costs of health services, practice variations, andinappropriate or useless interventions (1). The structure,process, and outcomes of care have all been the focus ofinvestigation, using either implicit or explicit criteria (2, 3).Although outcomes are the ultimate end results by which thevalue of medical interventions can be assessed, they are alsoconsidered less sensitive and sometimes even irrelevant inthis context (2, 3). This is certainly true with regard to breastcancer, where some major therapeutic innovations, such asthe shift toward locoregional management with breastpreservation, have not necessarily led to improvement insuch an important outcome of treatment as survival (4).

In 1985, Donabedian (5) described the "epidemiology ofquality" and the main predictors of optimal care among char-acteristics of the client, the attending physician, and the hos-pital of treatment. Using the same conceptual framework, thecurrent review was undertaken to identify variables associ-ated with the pattern of care offered to women with breastcancer, in an era where important therapeutic developmentshave taken place.

Our ultimate goal was to assess outputs as well as outcomesof treatment for breast cancer and their determinants and topromote evidence-based decision making in the organizationof health services in this area. This review should provideclues to be used by health planners and public health practi-tioners for breast cancer control, while also taking into accountthese patients' concerns and expectations. The abundance ofthe literature on health services in breast cancer and the uncer-tainty about the optimal treatment settings to be implementedboth underscore the relevance of this critical appraisal.

Received for publication May 3, 1999, and accepted for publica-tion May 5, 2000.

Abbreviations: Cl, confidence interval; OR, odds ratio; RR, relativerisk.

1 Direction de la Sante Publique, Regie Regionale de la Sante etdes Services Sociaux de Montreal-Centre, Montreal, Canada.

2Groupe de Recherche en Epidemiologie de I'Universite Lavaland Centre des Maladies du Sein, Hopital du Saint-Sacrement,Quebec, Canada.

3 Centre de Recherche, Centre Hospitalier de I'Universite deMontreal, Hopital Notre-Dame, Montreal, Canada.

Reprint requests to Dr. Nicole Hebert-Croteau, Direction de SantePublique de Montreal-Centre, Pavilion Lafontaine, 1301, rueSherbrooke Est, Montreal, PQ, H2L 1M3, Canada (e-mail:[email protected]).

MATERIALS AND METHODS

Studies included in this review were identified by a com-puterized search of the MEDLINE database for the timeperiod from 1985 to March 1999. The key words used werebreast neoplasms, in combination with any of the following:audit, medical audit, hospitals, oncology service hospital, out-come and process assessment, mastectomy, radiotherapy,chemotherapy, hormonal therapy, combined modality ther-apy, treatment variation, quality of medical care, and qualityof health care. In addition, reference lists of papers so identi-fied were searched for additional publications on these topics.

We excluded self-reports of therapeutic activities obtainedby surveys or hypothetical scenarios of case management. Inaddition, studies describing temporal or geographic varia-tions without hypothesis testing about organizational fea-tures related to the provision of care, as well as ecologicanalyses based on aggregated data without measurement onindividuals, were excluded. Finally, economic evaluations(cost minimization, cost-effectiveness, cost-utility analyses,and so on) were not included. This review was restricted topapers published in either French or English. Studies corre-sponding to these criteria appear in references 6-67.

OVERVIEW OF PUBLISHED STUDIES ON QUALITY OFCARE IN BREAST CANCER

The majority of studies on the evaluation of the patterns ofcare for breast cancer were published in the 1990s, and pub-lications on this topic were comparatively rare before thisperiod. These evaluations were conducted on different studypopulations, including local/regional (8, 11, 13, 16, 17,21-23, 25, 33, 34, 40, 42, 45, 49, 55, 58, 59, 61, 63-65) ornational (6, 9, 12, 19, 29, 30, 38, 46, 47, 66) samples ofbreast cancer patients in the United States, as well as inAustralia (14, 15, 27, 56, 57), Canada (31, 43, 51, 52, 67),Italy (7, 10, 24, 26, 32, 37, 39), Japan (28), Norway (50),South Africa (18), and the United Kingdom (20, 35, 36, 41,44, 48, 53, 54, 60, 62). We therefore reviewed provision ofcare under several types of health systems.

The majority of studies evaluated the process of care (6-8,10-14, 16,17, 19-24, 27, 29-43,45-55,57-61, 64-67) and,less often, outcomes of treatment (9, 15, 18, 20, 25, 26, 28,29,33,36,40,41,44,50, 55,56,60, 62,63). In the first case,the focus was on locoregional management (7, 8, 11-14, 16,19-24, 27, 29-43, 45-55, 57-61, 64-67), systemic adjuvanttreatment (7, 8, 10, 16, 17, 20, 22, 23, 27, 29, 35, 36, 40, 41,49, 50, 52-54, 59, 67), diagnostic procedures, such as stag-

228

Downloaded from https://academic.oup.com/epirev/article-abstract/22/2/228/457000by gueston 14 February 2018

Organizational Factors Related to Care for Breast Cancer 229

ing or measurement of specific tumor characteristics (e.g.,hormone receptors) (6-8, 12, 16, 20, 22, 27, 29, 40, 41, 49,50, 53, 59, 64, 66), and rehabilitation therapy (12, 30).Studies on outcomes estimated survival, and less frequentlyrecurrences and disease-free survival, at varying periods offollow-up. In one case (18), the focus was on complicationsof surgery. In a few publications, the study populations wererestricted to women younger than 50 years (41), above age50 (8, 54, 67), or above age 65 (19, 46, 47, 49, 55, 66).

Some adjustment for potential confounders was usuallyperformed. Among adjustment variables, the most frequentwere demographic characteristics, especially age at diagno-sis and some measure of socioeconomic status (education,income, occupation, or some other), as well as stage. In aminority of papers, extensive adjustment for case mix,defined as controlling for demographic, as well as disease,characteristics and comorbidity, was performed (8, 17, 25,32, 37, 38, 46, 55, 58, 59, 61, 63, 66, 67). A few studiesreported only crude unadjusted results (7, 13, 18, 27, 29, 35,60, 64).

Most of the time, hospital characteristics (type of hospital(health maintenance organization vs. other, CommunityClinical Oncology Program vs. other, affiliation or not withspecific cancer task force, cancer center vs. other, ownership,profit status), teaching status, bed size, cancer or breast can-cer caseload, rural vs. urban location, oncology charges,length of stay for patients admitted in oncology, percentageof patients with Medicare or Medicaid, research activities,health board, and on-site facilities related to cancer treat-ment) were investigated. In addition, some reports assessedvariables related to the client (especially insurance coverage)or the attending physician (age, sex, breast cancer caseload,board certification, experience with breast cancer, researchactivities, specialty or specialization in breast cancer treat-ment, year of graduation, university affiliation, physician-to-population ratio in the area). Some of these variables are infact combinations of others or are strongly correlated witheach other. This is the case, for example, of individual expe-rience with breast cancer, a combination of both years inpractice and usually year of graduation, and volume ofpatients treated by a given clinician. Similarly, physiciansspecialized in breast cancer are expected to see a larger vol-ume of cases. In addition, the surgeon's breast cancer case-load is also related, to some extent, to the volume of patientsadmitted into his/her hospital of practice. Adequate analyti-cal strategies should therefore attempt to disentanglecollinearity among these independent predictors, while alsoperforming adequate adjustment for individual differencesamong cases in various treatment settings.

CHARACTERISTICS OF THE CLIENT

We did not include in this review demographic character-istics, such as age, race, or socioeconomic status, that mightinfluence the treatment received but that do not truly refer tothe organization of health services. These were taken intoaccount as adjustment variables, but not for their main effectas predictors of care. Therefore, only insurance coverage orpayment status was considered in this review. Several stud-

ies assessed the influence of this factor on either the process(8, 12, 14, 16, 29, 30, 33, 38, 40, 42, 45, 55, 57-59, 61, 64,66) or outcomes (25, 29, 33, 40, 55) of treatment for breastcancer. Among these, six publications specifically comparedquality of care between members of health maintenanceorganizations and individuals with a different type of insur-ance coverage (33, 40, 55, 58, 59, 66). Managed care refersto an organization of services that extends beyond the soleissue of funding. These reports were, however, summarizedin this section but could as well have been included with thereview of hospital variables associated with care received.

Several studies suggested that extensive insurance cover-age is associated with more appropriate treatment for breastcancer (12, 29, 38, 42, 45, 64) and also with better survival(25). Some (12, 16, 25, 29) also evaluated the stage at pre-sentation or specific diagnostic procedures and found a sim-ilar favorable influence of more coverage. In the largeststudy conducted on care for breast cancer, Johantgen et al.(38) used data from the Hospital Cost and Utilization Projectto assess surgical treatment for breast cancer among 87,449cases diagnosed in the United States between 1981 and 1987.Self-expected payers had significantly less breast-conservingsurgery (odds ratio (OR) = 0.78, 95 percent confidenceinterval (CI): 0.68, 0.89), adjusted for age, stage, race,comorbidity, hospital region, location, bed size, and teachingstatus. This finding has also been shown in other settings aswell. Young et al. (42) looked at the cases of 9,288 womentreated for breast cancer in Pennsylvania between 1986 and1990 and found that those with private insurance were morelikely to receive breast-conserving surgery (51.1 percentwith Blue Cross or Blue Shield coverage vs. 43.6 percentwith Medicaid and 36.9 percent with Medicare, p < 0.05) andalso to subsequently have radiation therapy (79.9 percent,60.0 percent, and 80.6 percent, respectively; p < 0.001). Twoother studies also found greater proportions of breast-conserving surgery with more insurance coverage: the firstone by Kotwall et al. (45) on 20,760 patients in 157 hospitalsin North Carolina between 1988 and 1993 (OR of breast-conserving surgery with private insurance = 1.2, 95 percentCI: 1.0, 1.4, adjusted for age, residence, hospital caseload,and year of surgeon's graduation) and the second one byArnold et al. (64) on 238 cases admitted into a single teach-ing hospital in Jacksonville, Florida (40 percent breast-con-serving surgery in funded patients vs. 29 percent in others,p = 0.04). A further investigation by Kotwall et al. (61)found a higher probability of breast-conserving surgeryamong self-pay/Medicaid/Medicare patients (OR = 2.4, 95percent CI: 1.4, 4.2), as opposed to those with private insur-ance, but this study was conducted in a single regional med-ical center in North Carolina on a relatively small sample of634 cases. Although surgery with breast preservation doesnot necessarily represent better care for breast cancer, theneed for additional radiation therapy and the requirement forresources of the whole procedure probably explain why thistreatment is more readily available to women having moreextensive insurance coverage.

By contrast, a few investigations failed to show any rela-tion between the pattern of care and insurance status (8, 14,30, 57). In two of them, appropriateness of care was defined

Epidemiol Rev Vol. 22, No. 2, 2000


230 Hebert-Croteau et al.

as either a management score obtained by contrasting clini-cal and treatment data (8) or a two-step surgical procedure,not type of surgery per se (30). The other two studies (14,57), both conducted in Australia, did adjust to some extentfor differences in case mix, and negative findings with regardto patterns of surgical treatment could reflect local differ-ences in the organization of care, as well as cultural differ-ences, compared with the situation in North America.

Assessment of care in health maintenance organizations,for its part, suggests that quality of care for breast canceramong members of health maintenance organizations is atleast equivalent to care provided in fee-for-service settings.Lee-Feldstein et al. (33) compared patients admitted inhealth maintenance organizations and community hospitalsin Orange County, California, between 1984 and 1990 andfound better survival rates in large community hospitals (rel-ative risk (RR) of death at 5 years of follow-up = 0.74, 95percent CI: 0.59,0.94) and significantly worse rates in healthmaintenance organizations (RR = 1.63, 95 percent CI: 1.16,2.30) than in small community hospitals, after adjusting forage at diagnosis, tumor size, number of involved nodes, andsurgical treatment received. This contrasts with the majorityof studies on this issue, including two on Medicare benefi-ciaries by Potosky et al. (55) and Riley et al. (66), thatshowed greater likelihood of adequate locoregional manage-ment of women aged 65 years or older admitted in healthmaintenance organizations. Three other publications byGuadagnoli et al. (58,59) and Vernon et al. (40) found equiv-alent management of or survival after breast cancer in healthmaintenance organizations and other centers. All these stud-ies extensively adjusted for differences in case mix, butcomorbidity was not taken into account in two of them (33,40).

Altogether, studies on payment status and management ofbreast cancer suggest that women having more protection aremore likely to receive appropriate diagnostic work-up, aswell as less invasive procedures and treatment consistentwith current standards of practice.

CHARACTERISTICS OF THE ATTENDING PHYSICIAN

The association between individual characteristics of theattending physician and treatment or outcomes of breast can-cer has been assessed repeatedly. The most consistent findingis that better care is offered by clinicians having more expe-rience with this disease, either because they are specializedin treating it (35,44,52,65), have more years in practice (30,59, 67), or see a larger volume of cases each year (21, 36,48,56, 60). Table 1 summarizes the main features of studies onthe annual caseload of breast cancer patients of physiciansand the care provided.

In one study conducted by Ferguson et al. (13) in a singlemedical center in Atlanta, Georgia, surgery with breastpreservation and radiotherapy was less likely when womenwith stage I and II disease were treated by surgeons whocared for five or more cases each year, compared with thosewho had a smaller caseload (0-20 percent vs. 35 percent ofcases, p = 0.001). No further adjustment was performed.Two other studies found an individual effect of the physician

not explained by any personal attribute (43, 54). Moritz et al.(54) evaluated 600 women aged 49-79 years from the SouthEast Thames region with invasive tumor less than 2 cm andfound that surgeons operating on more than 20 such cases ina year had a lower mastectomy rate, although this differencedisappeared after adjustment for case mix. Strong individualvariation among physicians was observed, however, thatcould not be explained by measured variables. The secondstudy was conducted by Hislop et al. (43) on all 942 node-negative cases diagnosed in British Columbia in 1991 andfound a strong surgeon effect on the probability of breast-conserving surgery not explained by gender, year of gradua-tion, academic affiliation, and caseload, after adjustment forpatient and tumor characteristics. This contrasts with twosubsequent publications by the same group that showed asignificant association between the surgeon's year of gradu-ation and the probability of breast-conserving surgery incrude analysis (from 56.7 percent to 33.3 percent for sur-geons who graduated in 1980 or later to 1950 or earlier, over-all p for trend = 0.006) (51), as well as higher compliancewith guidelines for systemic therapy among women referredto an oncology specialist (52), after taking into account thetype of hospital (cancer center vs. community) and patientcharacteristics. This underscores the importance of properadjustment for all relevant potential confounders. Finally,four additional investigations, three of them populationbased (11, 27, 62), concluded that the process or outcomes ofcare for breast cancer do not substantially vary with theattending physician (11, 18, 27, 62), but none of them didsimultaneously adjust for characteristics of the hospital oftreatment.

The most reliable evidence about the relation between theexperience of the physician and care for breast cancer comesfrom those few investigations that simultaneously consideredcharacteristics of the patient, her physician, and the hospitalof treatment in multivariate analytical models. All such stud-ies did confirm that management of the disease variesaccording to the attending clinician. Among these, Satarianoet al. (21) used data on 2,238 patients from the metropolitanDetroit area Surveillance, Epidemiology, and End Resultsprogram database in 1985-1987 to study the associationbetween the clinician caseload and locoregional managementof breast cancer, while adjusting for hospital size and thepatient's age, race, marital status, and year of admission.Surgeons with a larger caseload were more likely to performbreast-conserving surgery with radiation therapy (OR =1.28, 95 percent CI: 1.03, 1.58 for those with >50 vs. <50cases each year). Hynes (30) estimated the influence of thephysician's experience and research activities, while control-ling for insurance coverage, size, type of hospital(Community Clinical Oncology Program vs. other commu-nity), and case mix as reflected by age, race, and nodal sta-tus, on the probability of two-step surgical procedures andrehabilitation therapy in 3,972 women with breast canceracross the United States selected from the National CancerInstitute Community Cancer Care Evaluation. Experiencedphysicians were more likely to perform two-step surgery(OR = 1.01, 95 percent CI: 1.00, 1.03) but also less likely toorder rehabilitation therapy (OR = 0.97,95 percent CI: 0.96,



Epid

em

iol

Rev

Vo

l.2

K3

o. 2, 2001

TABLE 1. Studies on physician

Author (ref. no.)

Mannetal. (11)

Ferguson et al. (13)

Satarianoet al. (21)

Byrne et al. (27)

Sainsbury et al. (36)

Hislop et al. (43)

Dixon et al. (48)

caseload

Year

1988

1990

1992

1993

1995

1996

1996

and process or outcomes

Location(no. of cases)

New Mexico, United States(n = 6,030)

Atlanta, United States(n= 183)

Detroit, United States(n = 2,238)

Western Australia(n = 692)

Yorkshire, United Kingdom(n= 12,861)

British Columbia, Canada(n = 942)

Edinburgh, United Kingdom

of care for breast cancer

Aspects ofcare studied

Surgical treatmentRadiation therapy



Diagnostic investigationsSurgical treatmentRadiation therapySystemic adjuvant therapy

Surgical treatmentRadiation therapySystemic adjuvant therapySurvival

Surgical treatment

Surgical treatment

Adjustmentvariables

Case mix (age, stage)

Case mix (age, race, maritalstatus), year of admission

Case mix (age, deprivation,extent of disease, tumorgrade)

Case mix (age, residence,income, tumor size, histology,location, extent of DCIS*)

Case mix (tumor size, nodal

Significantfindings

No association of BCS* and radiationtherapy with surgeon caseload

BCS with radiation therapy less likely withsurgeons' having a larger caseload

BCS with radiation therapy more likelywith surgeons' having a largercaseload

No association of BCS with surgeoncaseload

Combined therapy and survival morelikely with surgeons' having a highercaseload

Physician effect not explained bymeasured variables

Completeness of excision and BCS

Olivotto et al. (52)

Moritz et al. (54)

Stevenson et al. (56)

Goy et al. (60)

Twelves et al. (62)

(n=152)

1997 British Columbia, Canada(n = 661)

1997 South East Thames, UnitedKingdom (n = 600)

1997 Western Australia (n = 692)

1998 South East Thames, UnitedKingdom (n = 1,029)

1998 Scotland, United Kingdom(n = 2,148)

Radiation therapySystemic adjuvant therapyState-of-the-art treatment

Surgical treatmentRadiation therapySystemic adjuvant therapy

Survival

Radiation therapySurvival

Survival

status, type, mvasiveness,mammographic appearance,EIC,* localization, volumeof excision)

Case mix (age, residence,income, tumor size, margins,type, HR,* grade, location,focality, LVNI,* extent ofDCIS, no. of nodes sampled)

Case mix (tumor size, nodal status,grade)

Case mix (stage)

Case mix (age, deprivation, tumorsize, nodal status, stage, HR,treatment)

more likely with surgeons' having ahigher caseload

Referral to a specialist oncologistdecreased with surgeon caseload

Treatment not associated with caseloadPhysician effect not explained by measured

variables

Survival better with surgeons' having a highercaseload

Radiation therapy more likely with surgeons'having a higher caseload

Survival not associated with surgeon caseload

' BCS, breast-conserving surgery; DCIS, ductal carcinoma in situ; EIC, extensive intraductal component; HR, hormone receptors; LVNI, lymphatic, vascular, or neural invasion.

CQCD

N'S.o'QJ.

o533D22.a>

I983"CDCD0)en

O

8rou



0.99). Kotwall et al. (45), in a statewide analysis of all casesof primary breast cancer in North Carolina between 1988 and1993, observed that recently graduated surgeons were morelikely to perform breast-conserving surgery (OR = 1.6, 95percent CI: 1.2, 2.0), independently of the patient's age andinsurance status and of the hospital bed size. Guadagnoli etal. (58,59) compared the management of stage I and II breastcancer in Massachusetts and Minnesota during the period1993-1995 and evaluated the association of locoregional andsystemic treatment with insurance coverage, hospital charac-teristics (including on-site availability of the specializedresources needed for cancer treatment), and surgeon's gen-der, year of graduation, and board certification. Extensiveadjustment was done for demographic and disease character-istics and comorbidity. Although use of breast-conservingsurgery was not related to board certification or years sincegraduation, surgeons with more experience were more likelyto prescribe radiation therapy after breast-conserving surgeryin Massachusetts (OR = 3.8, 95 percent CI: 1.2, 12.0 forthose with >30 vs. <10 years since graduation) but not inMinnesota (OR = 1.3, 95 percent CI: 0.2, 10.7). Finally,Hebert-Croteau et al. (67) also focused on the managementof stage I and II breast cancer in Quebec, a region with a uni-versal free-of-charge health care system, during the period1988-1994, and studied the association of locoregional treat-ment with the time since graduation of the attending clini-cian, hospital breast cancer caseload, and participation inmulticenter clinical trials, adjusting for case mix (age at diag-nosis, characteristics of the disease, and comorbidity).Surgeons with 30 or more years in practice were twice aslikely to provide treatment consistent with consensus recom-mendations after adjustment for other covariates (OR = 2.0,95 percent CI: 1.4,2.9).

Overall, these data suggest that compliance with practiceguidelines, an indicator of quality of care, tends to increasewith the experience of the attending physician. In addition,breast-conserving surgery was more likely with younger ormore recently graduated physicians in several investigations(45, 51, 61, 65); this suggests that therapeutic innovationsfind easier application when they are learned through a basiceducational curriculum than when learned afterward. Noclear threshold for a desirable physician breast cancer case-load can be identified from these data, the numbers of 5(13), 20 (48, 56), 30 (36), and 50 (21) cases per year havingbeen reported as significantly associated with the patterns oroutcomes of treatment.

CHARACTERISTICS OF THE HOSPITAL OF TREATMENT

Among all the organizational aspects of care for breastcancer, the characteristics of the hospital of treatment havebeen the most extensively investigated. As for the physiciancaseload, greater bed size of the treatment center and, moreimportantly, cancer or breast cancer annual caseload havebeen shown to favorably influence the process of care (7, 16,21,24,38,45,47,53,59) and especially to increase die prob-ability of surgery with breast preservation (21,24, 38,45,47)and, in some instances, survival (33, 63). Studies on thisissue are listed in table 2.

A few publications describe the performance of Italianhospitals. The Interdisciplinary Group for Cancer CareEvaluation (7) studied 1,262 newly diagnosed cases ofbreast cancer admitted into 63 general hospitals in1983-1984. The relation between diagnostic investigationsand staging, surgical treatment, radiation therapy, systemicadjuvant therapy, and state-of-the-art treatment with type ofhospital (member or not of the Italian Breast Cancer TaskForce), hospital size, and on-site radiotherapy, oncology,and pathology showed that staging was more likely in largehospitals (82 percent vs. 76 percent, p < 0.05) and in thosewith oncology facilities (81 percent vs. 66 percent, p <0.001). On-site availability of radiation facilities influencedthe probability of a woman's receiving radiotherapy. Inaddition, breast-conserving surgery in younger women wasmore likely in hospital members of the Italian Breast CancerTask Force and with on-site oncology departments. In a sub-sequent publication by the same group on a subset of 353patients, no difference in the frequency of chemotherapycould be detected when hospitals were grouped according tosize, although the likelihood of receiving this treatment wasgreater in those members of the Italian Breast Cancer TaskForce (10). Grilli et al. (24) used the original database andevaluated if inappropriate Halsted mastectomy varied withvolume across hospitals. Inappropriate interventions wereless likely to be performed in centers with a large volume ofcases (OR = 0.36, 95 percent CI: 0.23, 0.55 for those with>20 vs. <10 patients each year, adjusted for education,tumor size, and nodal status).

Several studies in the United States have also confirmedthe association between the hospital bed size or breast cancercaseload and the pattern of care for breast cancer. In particu-lar, Johantgen et al. (38) found a positive association ofbreast-conserving surgery with hospital size, the odds ratiosin hospitals with <100 and 100-249 beds being 0.90 (95 per-cent CI: 0.83, 0.98) and 0.95 (95 percent CI: 0.89, 1.01),respectively, as compared with >250 beds. These estimatesare adjusted for age, extent of disease, comorbidity, paymentstatus, race, hospital location, teaching status, and region. Asimilar observation was made by Nattinger et al. (19, 47) on81,762 Medicare beneficiaries admitted in 1986 or 1990 into3,632 hospitals across the United States. Although bed sizewas not an independent predictor of breast-conservingsurgery in 1986 (19), the time trend showed that the percent-ages of this intervention in hospitals with <125 beds,126-300 beds, and 301-400 beds were 10.0 percent, 13.7percent, and 15.6 percent in 1986 and 9.5 percent, 14.2 per-cent, and 17.3 percent in 1990 (p values for comparison of1986 and 1990 = 0.40, 0.29, and <0.001, respectively).When the volume of breast cancer operations was the pre-dictor of interest, the percentages of breast-conservingsurgery in centers with <13, 13-23, and >24 cases per yearwere 11.9 percent, 14.2 percent, and 16.6 percent in 1986 and11.9 percent, 15.5 percent, and 19.3 percent in 1990 (p val-ues for comparison of 1986 and 1990 = 0.84, 0.007, and<0.001, respectively).

In addition, two odier investigations (33, 63) showed thatsurvival is better for women treated in larger hospitals. Thefirst study by Lee-Feldstein et al. (33) was conducted in




California over the period from 1984 to 1990 on 5,892 casesadmitted into 126 hospitals. This study showed better sur-vival in large compared with small community centers (RRof dying = 0.74, 95 percent CI: 0.59, 0.94 for localized dis-ease; RR = 0.74, 95 percent CI: 0.60, 0.91 for regional dis-ease, adjusted for age, tumor size, histology, nodal extension,and locoregional treatment). The second investigation byRoohan et al. (63) used the New York State hospital dis-charge database between 1984 and 1989 and found that treat-ment in very low volume hospitals was associated with a 60percent greater risk of all-cause mortality (RR of dying =1.6, 95 percent CI: 1.42, 1.81 for very low- vs. high-volumecenters), after adjusting for the type of surgery and thepatient's age, stage, comorbidity, race, socioeconomic status,and distance to hospital.

Taken together, these studies on hospital volume lead totwo main observations. Some of them suggest that individu-als admitted into very small hospitals are more likely to haveadverse features of their disease at presentation, to receivesuboptimal treatment, especially omission of radiation ther-apy or inadequate assessment of the status of the axilla, andto experience adverse outcomes. In addition, several studieshave shown a trend for several indicators of quality of care,including better survival, to increase with hospital caseload,although a leveling off of this effect beyond a minimalthreshold cannot be clearly ruled out.

Studies on this issue are not, however, unanimous. Severalreports have found no independent association (8, 15, 19, 26,28, 31, 49) or even a negative association (9, 12, 30, 37, 39,67) between hospital caseload and the probability of conserv-ative locoregional treatment, state-of-the-art management, orsurvival. It is unclear why the hospital bed size or caseloaddoes not consistently emerge as a significant predictor of carereceived or survival. Several of these investigations were pop-ulation based, therefore minimizing potential selection biasresulting from specific referral patterns, and were conductedon very large study populations. Although most did performsome adjustment for individual characteristics of patients,comorbidity was rarely taken into account. In a minority ofstudies (12, 21, 30, 45, 58, 59, 67), characteristics of thephysician responsible for treatment were also considered.Differences in the categorization of caseload and in the defi-nition of high- and low-volume centers might also explainsome of the discrepancies in results across studies. Finally,other aspects of the organization of services in hospitals, oftencorrelated with the volume of cases, could have been opera-tional but were not systematically considered. For example,large hospitals usually have an academic affiliation, andteaching status itself has been associated with a greater prob-ability of breast-conserving surgery or adequate locoregionalmanagement (12, 19, 20, 23, 33, 34, 37, 38, 41, 46, 51, 58),with better systemic treatment (17, 20, 23), or even in somecases with better survival (20,28). This is not unique to breastcancer and has been shown for several other health conditionsas well (68, 69). In addition, better quality of care, includinggreater access to surgery with breast preservation, has beenreported in urban centers (19, 38, 39, 42, 57), although notunanimously so (9, 16, 22, 49). On-site availability of spe-cialized equipment sometimes has been associated with better

care, again including greater likelihood of breast-conservingsurgery (7, 12, 19, 37, 46, 47). Although little difference hasbeen shown according to the type of hospital in several stud-ies (15, 26, 30, 47), a few investigations have suggested thatthe process of care is better in cancer centers (50, 52), in cen-ters with participation in breast cancer control activities (7,10), and in private hospitals (12, 17, 39). All of these charac-teristics usually reflect a larger proficiency of resources in agiven environment and could account, at least to some extent,for the association of hospital volume with the process or out-comes of treatment.

CLUES FOR RESEARCHERS AND PLANNERS OFHEALTH SERVICES

What conclusions can then be reached about the organiza-tion of health services most likely to result in optimal careand offering the best chances of survival to women withbreast cancer? Several of the associations reported here forbreast cancer had also been observed by Donabedian (5)almost two decades ago in his general overview. It is betterrecognized today that factors associated with the process oroutcomes of treatment tend to vary across different healthconditions or procedures (70, 71).

Although the literature is not consistent, there are severalindications that a larger caseload promotes the adoption oftechnologic or therapeutic innovations by clinicians and treat-ment centers, such as breast-conserving surgery or use of sys-temic adjuvant treatment. This, however, is not systematicallyequivalent to better treatment. For example, it has been repeat-edly demonstrated that surgery with breast preservation tendsto be used more frequently in larger centers (21, 24, 38, 45,47). With regard to consensus recommendations for combin-ing this surgery with radiation therapy and dissection of theaxilla (72), however, the available evidence sometimes led toan opposite conclusion (67). Studies aimed at evaluating indi-cations for and appropriateness of treatments could thereforeprovide a different account of the association between volumeand quality of care for breast cancer. In addition, analyticalissues, especially the use of several categorizations for case-load, while adjusting for other physician and hospital charac-teristics, need to be properly addressed (73). Finally, as hasbeen emphasized (71, 73), the association between volumeand patterns of care could in fact result from selective referralof individuals with favorable prognostic features in larger cen-ters and not from better performance with increasing practice.This alternative explanation seems unlikely, given that manystudies included in this review adjusted, at least partially, fordifferences in case mix.

Our interpretation of the literature and its policy implica-tions needs to be cautious. First, it is obvious from thisoverview that an important volume of research on quality ofcare for breast cancer, and for other health problems as well(74), suffers from methodological weaknesses, especiallyfailure to take into account all significant potential con-founders. Because virtually all studies in this area are obser-vational, improper adjustment for differences in case mix,the definition of which is closely dependent on the aspect ofcare studied (75), has been a limitation in several investiga-



TABLE 2. Studies on hospital volume and process or outcomes of care for breast cancerw

(TCD

63(D

2.01

Author (ref. no.) YearLocation

(no. of cases)Aspects of

care studiedAdjustmentvariables

Significantfindings

GIVIO* (7)

Greenfield et al. (8)

Mettlin et al. (9)

GIVIO(10)

Diehretal. (12)

Bonettetal. (15)

Handetal. (16)

15;

Rev

Vol. 22, N

o.2,2

0

Nattinger et al. (19)

Satariano et al. (21)

Grilli et al. (24)

Boffetta et al. (26)

Tanaka et al. (28)

Hynes (30)

Iscoe et al. (31)

1986 Italy (n = 1,262)

1987 Southern California, UnitedStates (n = 374)

Diagnostic investigationsStagingSurgical treatmentRadiation therapySystemic adjuvant therapyState-of-the-art treatment

Treatment score reflectingappropriate diagnosis,staging, and initial therapy

Staging more likely in large hospitals

Case mix (age, stage, comorbidity)

1987

1988

1989

1991

1991

1992

1992

1993

United States (n = 27,442)

Italy (n = 353)


South Australia (n = 2,589)

Chicago, United States(n = 5,766)


Detroit, United States(n = 2,238)

Northern and central Italy(n=985)

Survival

Systemic adjuvant therapy(chemotherapy)

Diagnostic investigationsHR assayRadiation therapyRehabilitation

Survival

Stage at diagnosisHR assayALN* dissectionRadiation therapySystemic adjuvant therapy

Surgical treatment


Surgical treatment

Case mix (age, race, stage,histology, HR,* treatment), sizeof metropolitan area

Case mix (nodal status)

Case mix (age, stage)

Case mix (age, tumor size, nodalstatus)

Case mix (age)

Case mix (race, nodal status),state, size of metropolitan area

Case mix (age, race, maritalstatus), year of admission

Case mix (age, marital status,occupation, education, tumor

1993 Piedmont, Italy (n = 4,764) Survival

1994 Osaka, Japan (n = 7,377) Survival

1994 United States {n = 3,972) Surgical treatmentRehabilitation

1994 Ontario, Canada (n = 12,815) Surgical treatment

size, nodal status, stage)

Case mix (age, marital status,occupation, area of birth), yearof diagnosis, town size

Case mix (age, stage, treatment)

Case mix (age, race, nodal status)

Case mix (age), county

Hospital variation not explained by differencesin bed size

Survival less likely in large hospitals

No independent association with hospitalbed size

Less appropriate care in large hospitals

No association of survival with hospital bedsize

Late stage and omission of radiation therapymore likely in hospitals with smallercaseloads


BCS* with radiation therapy more likely inlarge hospitals

Inappropriate Halsted procedure less likelyin large hospitals

No association of survival with hospitalcaseload


Two-step surgical procedure and rehabilitationless likely in large hospitals

BCS associated with individual hospitalpractices, not hospital caseload


I

ro

proroooo

Lee-Feldstein et al. (33)

Scorpiglione et al. (37)

Johantgen et al. (38)

Grill! and Repetto (39)

Kotwall et al. (45)

Nattinger et al. (47)

Hillner et al. (49)

Ma et al. (53)

Guadagnoli et al. (58)

Guadagnoli et al. (59)

Roohan et al. (63)

Hebert-Croteau et al. (67)

1994

1995

1995

1995

1996

1996

1996

1997

1998

California, United States(n = 5,892)

Italy (n= 1,724)


Lombardia, Italy (n = 6,961)

North Carolina, United States(n = 20,760)


Virginia, United States(n = 3,361)

North East Thames, UnitedKingdom (n = 309)

Massachusetts and Minnesota,United States (n = 2,575)

Surgical treatmentRadiation therapySurvival

Surgical treatment

Surgical treatment

Surgical treatment

Surgical treatment

Surgical treatment

Stage at diagnosisALN dissectionRadiation therapySystemic adjuvant therapy

Pathology report andpathologic staging

Surgical treatmentRadiation therapySystemic adjuvant therapyQuality-of-life factors (e.g.,

FNA)

Surgical treatment

Case mix (age, tumor size, nodalstatus, histology)

Case mix (age, stage, comorbidity]

Case mix (age, race, nodal status,comorbidity), region

Case mix (age, marital status,occupation), region, year ofadmission

Case mix (age, residence)

Size of metropolitan area

Case mix (age, race, marital statusresidence, income, education,comorbidity)

Case mix (age)

Case mix (age, marital status,residence, income, education,stage, comorbidity, pastcarcinomas), year of diagnosis

1998 Massachusetts and Minnesota, ALN dissectionUnited States (n = 2,575) Radiation therapy

Systemic adjuvant therapy

1998 New York State, United States Survival(n = 47,890)

1999 Quebec, Canada (n = 1,174) Surgical treatmentRadiation therapySystemic adjuvant therapyState-of-the-art treatment

Case mix (age, marital status,residence, income, education,stage, comorbidity, pastcarcinomas), year of diagnosis

Case mix (age, race, SES,* stage,comorbidity, distance tohospital, surgery)

Case mix (age, tumor size, grade,HR, comorbidity), year ofdiagnosis

Survival better in large community hospitals

Inappropriate Halsted procedures more likelyin large hospitals

BCS more likely in large hospitals

BCS more likely in small hospitals

BCS more likely in large hospitals

Change in BCS from 1986 to 1990 morelikely in large hospitals

bed size

Availability of FNA,* access to specialist sur-geon, and larger number of nodes excisedmore likely in hospitals with larger caseloads


Radiation therapy after BCS more likely inhospitals with >100 beds (Massachusetts)

Axillary dissection more likely in hospitals with£500 beds (Massachusetts)

Survival more likely in hospitals with largercaseloads

State-of-the-art treatment less likely inlarge hospitals

* GIVIO, Interdisciplinary Group for Cancer Care Evaluation; HR, hormone receptors; ALN, axillary lymph nodes; BCS, breast-conserving surgery; FNA, fine-needle aspiration; SES, socioeconomic status.

1o'

0)

aow3Jcp_wCD

a.oOQ>CDCD

CDCO

OCDOCD

rowui



tions. Attempts have been made to better characterize thestrength and direction of the resulting bias (76). Furtherassessment of the variables most critical to this end, as wellas their correlation and independent association with theprocess or outcomes of treatment, is needed.

Second, few evaluative studies have been performed thattook advantage of ongoing reforms and mergers of healthservices to confirm the influence of some organizational fea-tures of health services (74). These reforms have often beenundertaken following economic incentives to do so, underthe assumption that better efficiency and economies of scalewould result (74, 77). They could offer unique opportunitiesto confirm some relations, for example, with hospital vol-ume, in the context of longitudinal, controlled observations.

Finally, few attempts have been made to assess the impactof the organization of health services on such relevant out-comes of breast cancer as quality of life, recurrences, anddisease-free survival. Because locoregional management hasnot been shown to influence survival except in very specificsubgroups of cases (4, 78, 79), and given the recognizednecessity to follow women with carcinoma localized to thebreast for long periods of time in order to detect a survivaladvantage associated with specific therapeutic options (80),such studies are needed before firm conclusions can bereached about the organization of health services most suit-able for breast cancer control.

Several initiatives to improve quality of care to womenwith breast cancer have been undertaken in recent years (81).Among them, treatment guidelines and consensus develop-ment conferences have been used extensively (72, 80,82-87). Although their impact on actual medical practice hasoften been modest (88-95), some recent studies suggestedthat publication of guidelines, either as clinical alerts or for-mal sets of recommendations for treatment, has at times beenfollowed by appropriate changes in case management (96,97). This has been shown to be more likely when implemen-tation strategies using several sets of incentives have beenused (98, 99). For its part, state legislation to oblige disclo-sure of treatment options by physicians has had only a miti-gated success regarding the use of breast-conserving surgery(100).

In the United States, the implementation of the CommunityClinical Oncology Program after 1982 has been anothereffort, sponsored by the National Cancer Institute, to promotethe participation of community hospitals in collaborative clin-ical trials and to foster cancer prevention and control activi-ties over a broad scope of research projects (101-106). Thebasic principle underlying this and other similar initiatives,such as the cancer outreach programs of the large cancer cen-ters (107), is to offer access to state-of-the-art treatment whilemaintaining individuals as close to their home as possible.This concern, often expressed by individuals with cancer andtheir families, has been taken into account in the reform ofoncology services undertaken in several areas in recent years(108), and it should be incorporated in future evaluative stud-ies on quality of care for breast cancer.

Overall, more research is needed before decisions cansafely be made on the type and organization of services mostlikely to contribute to breast cancer control. This is an impor-

tant domain of investigation, likely to have major implica-tions for women afflicted with this disease, professionalsinvolved in their treatment, and health administrators as well.

ACKNOWLEDGMENTS

This review was performed in the context of an operatinggrant from the Fonds de la Recherche en Sante du Quebec.

REFERENCES

1. Blumenthal D. The origins of the quality-of-care debate. NEngl J Med 1996;335:1146-9.

2. Brook RH, McGlynn EA, Cleary PD. Measuring quality ofcare. N Engl J Med 1996;335:966-70.

3. Donabedian A. Evaluating the quality of medical care.Milbank Mem Fund Q 1966;44:166-203.

4. Early Breast Cancer Trialists' Collaborative Group. Effects ofradiotherapy and surgery in early breast cancer. N Engl J Med1995;333:1444-55.

5. Donabedian A. The epidemiology of quality. Inquiry 1985;22:282-92.

6. Mortenson LE, Kerner JF, Anderson PN, et al. Changes in thepatterns of patient work-up in community hospital oncologyprograms and in comparison hospitals. Prog Clin Biol Res1986;216:105-15.

7. Diagnosis and first-line treatment of breast cancer in Italiangeneral hospitals. G.I.V.I.O. (Interdisciplinary Group forCancer Care Evaluation)—Italy. Tumori 1986;72:273-83.

8. Greenfield S, Blanco DM, Elashoff RM, et al. Patterns of carerelated to age of breast cancer patients. JAMA 1987;257:2766-70.

9. Mettlin CJ, Schoenfeld ER, Natarajan N, et al. Interhospitaldifferences in cancer survival. J Chronic Dis 1987;40:157-64.

10. Performance of general hospitals in delivering adjuvant chemo-therapy to breast cancer patients. G.I.V.I.O. (InterdisciplinaryGroup for Cancer Care Evaluation), Italy. Tumori 1988;74:377-86.

11. Mann BA, Samet JM, Hunt WC, et al. Changing treatmentof breast cancer in New Mexico from 1969 through 1985.JAMA 1988;259:3413-17.

12. Diehr P, Yergan J, Chu J, et al. Treatment modality and qual-ity differences for black and white breast-cancer patientstreated in community hospitals. Med Care 1989,27:942-58.

13. Ferguson CM, Feinstein AC, Pendergrast WJ. Determinantsof primary therapy of early stage breast cancer. J Med AssocGa 1990;79:351^.

14. Hurley SF, Livingston PM, Jolley DJ, et al. Patterns of surgi-cal treatment of breast cancer in Victoria. Aust N Z J Surg1990;60:871-8.

15. Bonnet A, Roder D, Esterman A. Case-survival rates for infil-trating ductal carcinomas by category of hospital at diagnosisin South Australia. Med J Aust 1991;154:695-7.

16. Hand R, Sener S, Imperato J, et al. Hospital variables associ-ated with quality of care for breast cancer patients. JAMA1991;266:3429-32.

17. Schleifer SJ, Bhardwaj S, Lebovits A, et al. Predictors ofphysician nonadherence to chemotherapy regimens. Cancer1991 ;67:945-51.

18. Funnell IC, Crowe PJ, Dent DM. Does surgical experienceinfluence mastectomy complications? Ann R Coll Surg Engl1992;74:178-80.

19. Nattinger AB, Gottlieb MS, Veum J, et al. Geographic varia-tion in the use of breast-conserving treatment for breast can-




cer. N Engl J Med 1992;326:1102-7.20. Basnett I, Gill M, Tobias JS. Variations in breast cancer man-

agement between a teaching and a non-teaching district. EurJ Cancer 1992;28A: 1945-50.

21. Satariano ER, Swanson M, Moll PP. Nonclinical factors asso-ciated with surgery received for treatment of early-stagebreast cancer. Am J Public Health 1992;82:195-8.

22. Howe HL, Katterhagen JG, Yates J, et al. Urban-rural differ-ences in the management of breast cancer. Cancer CausesControl 1992;3:533-9.

23. Studnicki J, Schapira DV, Bradham DD, et al. Response to theNational Cancer Institute alert. The effect of practice guide-lines on two hospitals in the same medical community.Cancer 1993;72:2986-92.

24. Grilli R, Mainini F, Penna A, et al. Inappropriate Halsted mas-tectomy and patient volume in Italian hospitals. Am J PublicHealth 1993;83:1762-4.

25. Ayanian JZ, Kohler BA, Abe T, et al. The relation betweenhealth insurance coverage and clinical outcomes amongwomen with breast cancer. N Engl J Med 1993;329:326-31.

26. Boffetta P, Merletti F, Winkelmann R, et al. Survival of breastcancer patients from Piedmont, Italy. Cancer Causes Control1993;4:209-15.

27. Byrne MJ, Jamrozik K, Parsons RW, et al. Breast cancer inwestern Australia in 1989. II. Diagnosis and primary manage-ment. Aust N Z J Surg 1993;63:624-9.

28. Tanaka H, Hiyama T, Hanai A, et al. Interhospital differencesin cancer survival: magnitude and trend in 1975-1987 inOsaka, Japan. Jpn J Cancer Res 1994;85:680-5.

29. Osteen RT, Winchester DP, Hussey DH, et al. Insurance cov-erage of patients with breast cancer in the 1991 commissionon cancer patient care evaluation study. Ann Surg Oncol1994; 1:462-7.

30. Hynes DM. The quality of breast cancer care in local com-munities: implications for health care reform. Med Care 1994;32:328-40.

31. Iscoe NA, Goel V, Wu K, et al. Variation in breast cancersurgery in Ontario. Can Med Assoc J 1994;150:345-52.

32. Grilli R, Scorpiglione N, Nicolucci A, et al. Variation in useof breast surgery and characteristics of hospitals' surgicalstaff. Int J Qual Health Care 1994;6:233-8.

33. Lee-Feldstein A, Anton-Culver H, Feldstein PJ. Treatmentdifferences and other prognostic factors related to breast can-cer survival. Delivery systems and medical outcomes. JAMA1994;271:1163-8.

34. Foster RS, Farwell ME, Costanza MC. Breast-conservingsurgery for breast cancer: patterns of care in a geographicregion and estimation of potential applicability. Ann SurgOncol 1995,2:275-80.

35. Sainsbury R, Rider L, Smith A, et al. Does it matter whereyou live? Treatment variation for breast cancer in Yorkshire.Br J Cancer 1995;71:1275-8.

36. Sainsbury R, Haward B, Rider L, et al. Influence of clinicianworkload and patterns of treatment on survival from breastcancer. Lancet 1995;345:1265-70.

37. Scorpiglione N, Nicolucci A, Grilli R, et al. Appropriatenessand variation of surgical treatment of breast cancer in Italy:when excellence in clinical research does not match with gen-eralized good quality care. J Clin Epidemiol 1995;48:345-52.

38. Johantgen ME, Coffey RM, Harris R, et al. Treating early-stage breast cancer: hospital characteristics associated withbreast-conserving surgery. Am J Public Health 1995;85:1432-4.

39. Grilli R, Repetto F. Variation in use of breast-conservingsurgery in Lombardia, Italy. Int J Technol Assess Health Care1995;ll:733-40.

40. Vernon S W, Heckel V, Jackson GL. Medical outcomes of carefor breast cancer among health maintenance organization andfee-for-service patients. Clin Cancer Res 1995; 1:179-84.

41. Richards MA, Wolfe CDA, Tilling K, et al. Variation in themanagement and survival of women under 50 years withbreast cancer in the South East Thames region. Br J Cancer1996;73:751-7.

42. Young WW, Marks SM, Kohler SA, et al. Dissemination ofclinical results. Mastectomy versus lumpectomy and radiationtherapy. Med Care 1996;34:1003-17.

43. Hislop TG, Olivotto IA, Coldman AJ, et al. Variations inbreast conservation surgery for women with axillary lymphnode negative breast cancer in British Columbia. Can J PublicHealth 1996;87:390-4.

44. Gillis CR, Hole DJ. Survival outcome of care by specialistsurgeons in breast cancer: a study of 3,786 patients in the westof Scotland. BMJ 1996;312:145-8.

45. Kotwall CA, Covington DL, Rutledge R, et al. Patient, hospi-tal, and surgeon factors associated with breast conservationsurgery. A statewide analysis in North Carolina. Ann Surg1996;224:419-29.

46. Ballard-Barbash R, Potosky AL, Harlan LC, et al. Factorsassociated with surgical and radiation therapy for early stagebreast cancer in older women. J Natl Cancer Inst 1996;88:716-26.

47. Nattinger AB, Gottlieb MS, Hoffman RG, et al. Minimalincrease in use of breast-conserving surgery from 1986-1990.Med Care 1996;34:479-89.

48. Dixon JM, Ravisekar O, Cunningham M, et al. Factors affect-ing outcome of patients with impalpable breast cancerdetected by breast screening. Br J Surg 1996;83:997-1001.

49. Hillner BE, Penberthy L, Desch CE, et al. Variation in stagingand treatment of local and regional breast cancer in theelderly. Breast Cancer Res Treat 1996;40:75-86.

50. Raabe NK, Kaaresen R, Fossaa SD. Hospital-related differ-ences in breast cancer management. Breast Cancer Res Treat1997;43:225-35.

51. Goel V, Olivotto I, Hislop TG, et al. Patterns of initial man-agement of node-negative breast cancer in two Canadianprovinces. Can Med Assoc J 1997; 156:25-35.

52. Olivotto IA, Coldman AJ, Hislop TG, et al. Compliance withpractice guidelines for node-negative breast cancer. J ClinOncol 1997;15:216-22.

53. Ma M, Bell J, Campbell S, et al. Breast cancer management:is volume related to quality? Br J Cancer 1997;75:1652-9.

54. Moritz S, Bates T, Henderson SM, et al. Variation in manage-ment of small invasive breast cancers detected on screening inthe former South East Thames region: observational study.BMJ 1997;315:1266-72.

55. Potosky AL, Merrill RM, Riley GF, et al. Breast cancersurvival and treatment in health maintenance organizationand fee-for-service settings. J Natl Cancer Inst 1997;89:1683-91.

56. Stevenson JM, Bochenek P, Jamrozik K, et al. Breast cancerin western Australia at 5 years after diagnosis. Aust N Z JSurg 1997;67:250-5.

57. Adelson P, Lim K, Churches T, et al. Surgical treatment ofbreast cancer in New South Wales 1991, 1992. Aust N Z JSurg 1997;67:9-14.

58. Guadagnoli E, Weeks JC, Shapiro CL, et al. Use of breast-conserving surgery for treatment of stage I and stage II breastcancer. J Clin Oncol 1998;16:101-6.

59. Guadagnoli E, Shapiro CL, Weeks JC, et al. The quality ofcare for treatment of early stage breast carcinoma. Is it con-sistent with national guidelines? Cancer 1998;83:302-9.

60. Goy JCM, Dobbs HJ, Henderson S, et al. Variations in refer-ral pattern for postoperative radiotherapy of patients withscreen-detected breast cancer in the South Thames EastRegion. Clin Oncol 1998;10:24-9.

61. Kotwall C, Covington D, Churchill P, et al. Breast conserva-tion surgery for breast cancer at a regional medical center. AmJ Surg 1998;176:510-14.

62. Twelves CJ, Thomson CS, Gould A, et al. Variation in the sur-vival of women with breast cancer in Scotland. Br J Cancer1998;78:566-71.

63. Roohan PJ, Bickell NA, Baptiste MS, et al. Hospital volumedifferences and five-year survival from breast cancer. Am JPublic Health 1998,88:454-7.

64. Arnold RE, Frykberg ER, Kilkenny JW, et al. Trends in sur-




gical treatment of breast cancer at an urban teaching hospital:a six-year review. Am Surg 1998;64:107-11.

65. Blair SL, O'Shea KE, Orr RK. Surgeon variability in treatingnonpalpable breast cancer: surgical oncology as a value-added specialty. Ann Surg Oncol 1998;5:28-32.

66. Riley GF, Potosky AL, Klabunde CN, et al. Stage at diagnosisand treatment patterns among older women with breast can-cer—an HMO and fee-for-service comparison. JAMA 1999;281:720-6.

67. H6bert-Croteau N, Brisson J, Latreille C, et al. Compliancewith consensus recommendations for the treatment of earlystage breast carcinoma in elderly women. Cancer 1999;85:1104-13.

68. Taylor DH, Whellan DJ, Slogan FA. Effects of admission to ateaching hospital on the cost and quality of care for Medicarebeneficiaries. N Engl J Med 1999;340:293-9.

69. Kassirer JP. Hospitals, heal yourselves. N Engl J Med 1999;340:309-10.

70. Sowden AJ, Grilli R, Rice N. The relationship between hos-pital volume and quality of health outcomes. York, UK: NHSCentre for Reviews and Dissemination, University of York,1997. (CRD report 8, part I).

71. Luft HS, Hunt SS, Maerki SC. The volume-outcome relation-ship: practice-makes-perfect or selective-referral patterns?Health Serv Res 1987;22:157-82.

72. National Institutes of Health. Treatment of early-stage breastcancer. NIH Consensus Conference. JAMA 1991;265:391-5.

73. Luft HS, Garnick DW, Mark DH, et al. Theoretical andmethodological issues in the volume-outcome relationship. In:Hospital volume, physician volume, and patient outcomes.Assessing the evidence. Ann Arbor, MI: Health AdministrationPress Perspectives, 1990:9-30.

74. Hospital volume and health care outcomes, costs and patients'access. Effective Health Care 1996;2:1-16.

75. Jencks SF, Dobson A. Refining case-mix adjustment. N EnglJMed 1987;317:679-86.

76. Davenport RJ, Dennis MS, Warlow CP. Effect of correctingoutcome data for case mix: an example from stroke medicine.BMJ 1996;312:1503-5.

77. Ferguson B, Posnett J, Sheldon T. Concentration and choicein the provision of hospital services. York, UK: NHS Centrefor Reviews and Dissemination, University of York, 1997.(CRD report 8, summary).

78. Overgaard M, Hansen PS, Overgaard J, et al. Postoperativeradiotherapy in high-risk premenopausal women with breastcancer who receive adjuvant chemotherapy. N Engl J Med1997;337:949-55.

79. Ragaz J, Jackson SM, Le N, et al. Adjuvant radiotherapy andchemotherapy in node-positive premenopausal women withbreast cancer. N Engl J Med 1997;337:956-62.

80. Goldhirsch A, Glick JH, Gelber RD, et al. Meeting highlights:international consensus panel on the treatment of primarybreast cancer. J Natl Cancer Inst 1998;90:1601-8.

81. Rapport sur le Forum National sur le Cancer du Sein. (InFrench). Ottawa, Canada: Ministere des Approvisionnementset Services, Gouvemement du Canada, 1994.

82. NIH consensus conference: adjuvant therapy for breast can-cer. JAMA 1985;254:3461-3.

83. US National Cancer Institute. Clinical alert from the NationalCancer Institute. Breast Cancer Res Treat 1988;12:3-5.

84. Glick JH. Meeting highlights: adjuvant therapy for breast can-cer. J Natl Cancer Inst 1988,80:471-5.

85. Glick JH, Gelber RD, Goldrisch A, et al. Meeting highlights:adjuvant therapy for primary breast cancer. J Natl Cancer Inst1992;84:1479-85.

86. Goldhirsch A, Wood WC, Senn HJ, et al. Meeting highlights:international consensus panel on the treatment of primarybreast cancer. J Natl Cancer Inst 1995;87:1441-5.

87. Comite directeur des guides de pratique clinique pour la priseen charge et le traitement du cancer du sein. Guides de pra-tique clinique pour la prise en charge et le traitement du can-

cer du sein. Document de concertation canadien. (In French).J Assoc Med Can 1998;158(suppl):SF49-58.

88. Lomas J. Words without action? The production, dissemina-tion and impact of consensus recommendations. Annu RevPublic Health 1991;12:41-65.

89. Kassirer JP. The quality of care and the quality of measuringit. N Engl J Med 1993;329:1263-5.

90. Greco PJ, Eisenberg JM. Changing physicians' practices. NEngl J Med 1993;329:1271^.

91. Fineberg HV. Clinical evaluation: how does it influence med-ical practice? Bull Cancer 1987;74:333-46.

92. Grimshaw JM, Russell IT. Effect of clinical guidelines onmedical practice: a systematic review of rigorous evaluations.

- Lancet 1993;342:1317-22.93. Kosecoff J, Kanouse DE, Brook RH. Changing practice pat-

terns in the management of primary breast cancer: consensusdevelopment program. Health Serv Res 1990;25:809-23.

94. Grilli R, Apolone G, Marsoni S, et al. The impact of patientmanagement guidelines on the care of breast, colorectal, andovarian cancer patients in Italy. Med Care 1991;29:50-63.

95. Ford LG, Hunter CP, Diehr P, et al. Effects of patient man-agement guidelines on physician practice patterns: theCommunity Hospital Oncology Program experience. J ClinOncol 1987;5:504-ll.

96. Johnson TP Ford L, Warnecke RB, et al. Effect of a NationalCancer Institute clinical alert on breast cancer practice pat-terns. J Clin Oncol 1994;12:1783-8.

97. Sawka C, Olivotto I, Coldman A, et al. The associationbetween population-based treatment guidelines and adjuvanttherapy for node-negative breast cancer. Br J Cancer 1997;75:1534-42.

98. US Congress, Office of Technology Assessment. Identifyinghealth technologies that work: searching for evidence.Washington, DC: US Government Printing Office, 1994.(Publication no. OTA-H-608).

99. Chassin MR. Improving the quality of care. N Engl J Med1996;335:1060-3.

100. Nattinger AB, Hoffmann RG, Shapiro R, et al. The effect oflegislative requirements on the use of breast conservingsurgery. N Engl J Med 1996;335:1035^0.

101. Lazovich D, Mowry PA, Flam MS, et al. The management ofbreast cancer in a community: the impact of the CCOP inFresno, California. Prog Clin Biol Res 1987;248:217-24.

102. Kaluzny AD, Ricketts T, Warnecke R, et al. Evaluating orga-nizational design to assure technology transfer: the case of theCommunity Clinical Oncology Program. J Natl Cancer Inst1989;81:1717-25.

103. Schopler JH. Ecological analysis of the first generation ofcommunity clinical oncology programs. Health Serv Res1993;28:69-95.

104. Kaluzny AD, Lacey LM, Warnecke R, et al. Predicting theperformance of a strategic alliance: an analysis of theCommunity Clinical Oncology Program. Health Serv Res1993;28:159-82.

105. Kaluzny AD, Warnecke RB, Lacey LM, et al. Cancer preven-tion and control within the National Cancer Institute's clinicaltrials network: lessons from the Community ClinicalOncology Program. J Natl Cancer Inst 1993:85:1807-11.

106. Frelick RW. The Community Clinical Oncology Program(CCOP) story: review of community oncologists' experienceswith clinical research trials in cancer with an emphasis on theCCOP of the National Cancer Institute between 1982 and1987. J Clin Oncol 1994; 12:1718-23.

107. Smith TJ, Desch CE, Simonson CJ, et al. Teaching specialtycancer medicine in rural hospitals: the cancer outreachprogam as a model. J Cancer Educ 1991:6:235-40.

108. Comite' Consultatif sur le Cancer. Programme quebe'cois delutte contre le cancer. Pour lutter efficacement contre le can-cer, formons 6quipe. (In French). Quebec, Canada: Ministerede la Sante et des Services Sociaux, Gouvemement duQuebec, 1997.



review of organizational factors related to care offered to women

Documents