302
The Quality of Care for Treatment of Early Stage
Breast Carcinoma
Is It Consistent with National Guidelines?
Edward Guadagnoli, Ph.D.1
Charles L. Shapiro, M.D.2
Jane C. Weeks, M.D.2
Jerry H. Gurwitz, M.D.3
Catherine Borbas, Ph.D., M.P.H.4
Stephen B. Soumerai, Sc.D.5
1
Department of Health Care Policy, Harvard Med-
ical School, Boston, Massachusetts.
2
Medical Oncology, Dana-Farber Cancer Institute,
Boston, Massachusetts.
3 pausal women with positive lymph nodes, and hormonal therapy for postmeno-
Meyers Primary Care Institute, Worcester, Mas-
sachusetts.
4
Healthcare Education Research Foundation, St.
Paul, Minnesota.
5
Department of Ambulatory Care and Prevention,
Harvard Pilgrim Health Care, Boston, Massachu-
setts.
Supported by Grants (CA59408 and CA57755)
from the National Cancer Institute.
During the time this study was conducted Dr.
Guadagnoli was a Picker/Commonwealth Scholar
and Dr. Gurwitz was the recipient of a Clinical
Investigator Award (K08 AG00510) from the Na-
tional Institute on Aging.
The authors wish to thank Marilyn Sarnie, M.A.,
Shera Gruen, Kathyrn Cleghorn, Nora Morris, M.A.,
and Shirley Germann, R.N., for their help in con-
ducting the study and Robert Wolf, M.S., for sta-
tistical programming.
Address for reprints: Edward Guadagnoli, Ph.D.,
Department of Health Care Policy, Harvard Medical
School, 180 Longwood Avenue, Boston, MA
02115.
Received November 5, 1997; revision received
January 23, 1998; accepted January 23, 1998.
© 1998 American Cancer Society
BACKGROUND. In response to the importance of early stage breast carcinoma as a
public health concern and to the complexity of the clinical literature devoted to
treatment of the disease, the National Institutes of Health has held a series of
Consensus Development Conferences on the treatment of early stage breast car-
cinoma. The authors assessed compliance with standards of care for women
treated in two states.
METHODS. The authors identified patients diagnosed at 18 randomly selected
hospitals (N 5 1514) in Massachusetts and at 30 hospitals (N 5 1061) in Minnesota.
They collected data from medical records, patients, and their surgeons to assess
compliance with four indicators of quality of care: radiation therapy after breast-
conserving surgery, axillary lymph node dissection, chemotherapy for premeno-
pausal women with positive lymph nodes and positive estrogen receptor status.
RESULTS. Rates of compliance for 3 of the 4 standards of care were . 80% in both
states. Only the rate for hormonal therapy for postmenopausal women was low (,
64%). However, the proportion of these women who received either chemotherapy
or hormonal therapy was . 90% in both states.
CONCLUSIONS. In the states studied, practice appears to be consistent with the
results of national consensus conferences and clinical trials regarding the treat-
ment of early stage breast carcinoma. For practices demonstrated to be associated
definitively with better outcomes (for example, chemotherapy for premenopausal
women with positive lymph nodes) or to be important with respect to prognosis
(axillary lymph node dissection) high rates of compliance were observed. Cancer
1998;83:302–9. © 1998 American Cancer Society.
KEYWORDS: quality of care, early stage breast carcinoma, axillary lymph node
dissection, radiation therapy, adjuvant therapy, health services research.
I n response to the importance of early stage breast carcinoma as a
public health concern and to the complexity of the clinical literature
devoted to treatment of the disease, the National Institutes of Health
has held a series of Consensus Development Conferences on the
treatment of early stage breast carcinoma.1–5 These conferences, dat-
ing from 1979, have addressed the surgical management of localized
disease, the role of radiation therapy, and the value of adjuvant
therapy in various subgroups of patients. Elements of these recom-
mendations have been reinforced by other conferences and/or pub-
lications.6 – 8
Four standards of care for early stage breast carcinoma (TNM
Stages I and II) have been identified1– 8: 1) radiation therapy after
breast-conserving surgery, 2) axillary lymph node dissection, 3) che-

motherapy for premenopausal women with positive
lymph nodes, and 4) hormonal therapy for postmeno-
pausal women with positive lymph nodes and positive
estrogen receptor status. The issue of breast-conser-
vation therapy versus mastectomy has been addressed
by national guidelines2 and breast conservation was
identified as the preferred treatment. However, given
that survival is equivalent for the two procedures, the
type of surgery may not be a good indicator of quality
of care for women with early stage breast carcinoma.
Of more relevance to quality may be whether the
patient is informed of both options for surgical treat-
ment.9
Much of the information published regarding the
care of patients with early stage breast carcinoma has
addressed the issue of surgical treatment.10 –15 How-
ever, Hand et al.16 used 1988 cancer registry data from
Illinois to determine the degree of compliance with
clinical standards other than type of surgery. They
found that for mature clinical standards (hormone
receptor determination and axillary lymph node dis-
section) compliance was good (88%), but for stan-
dards introduced closer to 1988 (radiation therapy
after breast-conserving surgery and adjuvant therapy)
compliance was not as good (, 57%).
We assessed compliance with standards of care
for women diagnosed with early stage breast carci-
noma in Massachusetts (N 5 1514) and Minnesota
(N 5 1061), 2 states reported to differ with respect to
treatment of breast carcinoma.11 We also examined
whether patient, physician, and hospital characteris-
tics were related to compliance with standards to
identify potential targets for quality improvement. We
collected data from medical records, patients, and
physicians in both states.
METHODS
Sample
Massachusetts
We stratified hospitals by county, presence or absence
of a residency training program, and bed size; 20 hos-
pitals were randomly selected. Eighteen hospitals
agreed to participate in the study. We accrued patients
with Stage I or II disease (TNM Stages I, IIA, and IIB)
between September 1993 and September 1995 from
hospital pathology offices or from hospital tumor reg-
istries. We excluded patients for whom the recom-
mendations from the national consensus conferences
were not likely to apply: those with carcinoma in situ,
bilateral synchronous carcinoma, and inflammatory
carcinoma.
Breast Carcinoma Quality of Care/Guadagnoli et al. 303
Minnesota
Thirty hospitals from throughout the state of Minne-
sota that are part of a consortium formed by the
Healthcare Education and Research Foundation (St.
Paul, MN) participated in the study. We accrued pa-
tients between January and December 1993 and used
the same eligibility criteria described for the Massa-
chusetts sample.
Data Collection
Medical records
Experienced data abstractors in both states docu-
mented features of patients’ clinical status, diagnostic
workup, primary treatment, adjuvant treatment, and
demographic characteristics from medical records lo-
cated in hospitals or outpatient surgical centers. Stage
of disease was coded on the basis of clinical informa-
tion documented in the medical record when patho-
logic information was not available. Data collection
occurred a minimum of 2 months after surgery to
maximize the likelihood of documenting all treat-
ments received and the results of diagnostic and prog-
nostic tests. For each data collector, a randomly se-
lected sample of medical records was reabstracted by
an independent record reviewer. Agreement between
the data collectors and the independent reviewer was
high (. 95%).
Patient survey
Patients participated in a telephone survey an average
of 3–5 months after diagnosis. Patients were asked
whether radiation therapy and/or adjuvant drug ther-
apy were received or scheduled to be received.
Patient income and education
We obtained income and education from a secondary
source (United States Census data from 1990) because
all patients did not participate in the survey phase of
the project, and because this information often is
missing at a high rate for those who do answer sur-
veys. We defined income for each patient as the me-
dian household income associated with the zip code
of her residence17 and education as the proportion of
adults with a high school education in her zip code.18
Physician survey
We attempted to survey each patient’s surgeon by
mail within 1 month of diagnosis. For an individual
patient, surgeons indicated whether radiation therapy
and/or adjuvant drug therapy were received or sched-
uled to be received.
304 CANCER July 15, 1998 / Volume 83 / Number 2
Hospital and surgeon characteristics
We obtained bed size, teaching status, presence of a
cancer program approved by the American College of
Surgeons, and presence of a radiation facility for hos-
pitals in each state from the American Hospital Asso-
ciation.19
For surgeons from Minnesota, we obtained a data
set indicating year of graduation from medical school,
gender, and board certification from the Healthcare
Education and Research Foundation. Similar data for
surgeons from Massachusetts were obtained from a
data set purchased from Foilo Associates, Inc. (Boston,
MA).20
Definition of Variables
We determined whether each standard of care (radia-
tion therapy after breast-conserving surgery, axillary
lymph node dissection, chemotherapy for premeno-
pausal women with positive lymph nodes, and hor-
monal therapy for postmenopausal women with pos-
itive lymph nodes and positive estrogen receptor
status) occurred by scanning across data sources ac-
cording to a coding algorithm. For example, we deter-
mined whether radiation therapy followed breast-con-
serving surgery by first consulting the patient survey.
If a patient survey was not available, we looked for this
information in the physician survey; if a physician
survey was not available, we consulted the medical
record data base. If information regarding any quality
indicator was missing from all data sources, we re-
quested this information from the patient’s surgeon
(this was a direct request to surgeons independent of
the physician survey) and/or we requested it from the
tumor registry at the hospital in which the patient
received primary therapy.( A copy of the coding algo-
rithms for each variable is available from the authors
upon request.)
Analysis of Data
Describing Quality of Care
We described patient demographic characteristics and
case mix as well as surgeon and hospital characteris-
tics for each state. For each quality indicator, we re-
ported the rate of use and 95% confidence intervals
(95% CI) by state.
Assessing the Correlates of Quality of Care by State
We conducted our analyses by state to examine
whether the correlates for each standard of care were
consistent across states. For radiation therapy after
breast-conserving surgery and axillary lymph node
dissection, we assessed the influence of age, educa-
tion, income, membership in a health maintenance
organization, location of residence, gender of patient’s
surgeon, board certification of patient’s surgeon, years
since graduation of patient’s surgeon, teaching status
of hospital, size of hospital, presence of a cancer pro-
gram approved by the American College of Surgeons
at hospital, and presence of a radiation facility at
hospital using logistic regression analysis. Covariates
in the models included marital status, severity of co-
morbid disease,21 past history of breast carcinoma,
past history of other cancer, and stage of disease (not
included in the model for axillary lymph node dissec-
tion). Year of diagnosis was included in the models for
Massachusetts. All models met the goodness of fit
criteria suggested by Hosmer and Lemeshow.22 We did
not perform logistic regression analyses for the adju-
vant drug therapy variables because the number of
patients eligible for each analysis was too small.
Approval of the Study Protocol
The study protocol was approved by the institutional
review board of Harvard Medical School. The study
protocol also was reviewed by institutional review
boards at hospitals or, when an institutional review
board did not exist, by hospital administrators respon-
sible for reviewing participation in research projects.
Patients were not contacted for participation in the
survey without the permission of their surgeon.
RESULTS
Hospital Characteristics
Hospitals from Massachusetts that agreed to partici-
pate in the study did not differ from the population of
Massachusetts hospitals with respect to bed size, lo-
cation (urban vs. rural), teaching status, presence of a
radiation facility, or presence of a cancer program
approved by the American College of Surgeons. Be-
cause the hospitals from Minnesota were not ran-
domly selected to participate in the study, the match
between the sample and population did not coincide
as well as did the Massachusetts sample and popula-
tion. Participating hospitals from Minnesota were
more likely to be larger (P , 0.001), to be located in a
urban area (P , 0.001), to have a radiation facility (P 5
0.007), and to have a cancer program approved by the
American College of Surgeons (P 5 0.01) than the
population of hospitals in the state. However, hospi-
tals belonging to the consortium represented a large
proportion (. 60%) of all statewide admissions.
Sample Characteristics
The characteristics of women in each state are pre-
sented in Table 1. Most patients were age $ 60 years,
white, diagnosed with Stage I disease, and did not
have a past history of breast carcinoma or other can-
 Breast Carcinoma Quality of Care/Guadagnoli et al. 305

TABLE 1 cer. Most patients were treated by surgeons who were

Characteristics of Women Diagnosed with Early Stage Breast male, were Board-certified, and had $ 20 years of
Carcinoma According to State
experience.
State We conducted 796 interviews with patients in
Massachusetts and 774 interviews with patients in
Massachusetts Minnesota Minnesota. The response rate was 61% in Massachu-
Characteristic (N 5 1514)a (N 5 1061)a
setts and 84% in Minnesota after taking into account
Demographic women who missed the consent procedure, women
Age (yrs) No. (%) who had language difficulties, and women whom we
, 50 406 (27) 253 (24) were unable to locate. In Massachusetts, those inter-
50–59 340 (23) 227 (21)
60–69 322 (21) 246 (23)
viewed were more likely than those not interviewed to
70–79 303 (20) 220 (21) be younger (P 5 0.001), to be married (P 5 0.001), to
$ 80 143 (9) 115 (11) have higher incomes (P 5 0.001), to live in areas in
African-American No. (%) 56 (4) 15 (1) which more residents had a high school education
Married No. (%) 860 (57) 619 (58) (P 5 0.001), to be members of health maintenance
Median household incomeb No. (%)
,$40,000 814 (54) 709 (67)
organizations (P 5 0.001), to have less severe comor-
Percent in zip code with high school bid disease (P 5 0.001), to have no prior history of
education No. (%) cancer (P 5 0.001), and to be treated by younger
, 70% 195 (13) 31 (3) surgeons (P 5 0.02), at larger hospitals (P 5 0.02), at
70–80% 394 (26) 203 (19) teaching hospitals (P 5 0.001), at hospitals with radi-
80–90% 571 (38) 468 (44)
$ 90% 327 (22) 342 (32)
ation facilities (P 5 0.001), and at hospitals with ap-
Urban residence No. (%) 1433 (95) 963 (91) proved cancer programs (P 5 0.03). In Minnesota,
Health maintenance organization those interviewed were more likely than those not
member No. (%) 407 (27) 419 (40) interviewed to be younger (P 5 0.001) and married
Year of diagnosis No. (%) (P 5 0.001).
1993 205 (14) 1061 (100%)
1994 768 (51)
Surgeons returned 65% (Massachusetts) and 74%
1995 541 (36) (Minnesota) of the surveys mailed to them. Patients
Case mix from Massachusetts without a corresponding survey
Comorbid disease No. (%) from their surgeon were more likely than other pa-
None 562 (37) 355 (33) tients to be older (P , 0.001), to live in areas in which
Mild 290 (19) 258 (24)
Moderate 577 (38) 359 (34)
fewer residents had a high school education (P ,
Severe 68 (5) 89 (8) 0.001), to have more severe comorbid disease (P ,
Past history of breast carcinoma No. (%) 112 (7) 90 (9) 0.001), to be treated by older surgeons (P , 0.001), to
Past history of other carcinoma No. (%) 102 (7) 89 (8) be treated by male surgeons (P P , 0.001), at hospitals
Stage I disease No. (%) 868 (57) 619 (58) without a radiation facility (P , 0.001), and at non-
Surgeon characteristics
Years since graduation - N (SD)
teaching hospitals (P , 0.001). Patients from Minne-
, 10 67 (5) 98 (10) sota without a corresponding survey from their sur-
10–19 598 (40) 359 (36) geon were more likely than others to be treated by
20–29 431 (29) 291 (30) older surgeons (P , 0.05), to be treated by Board-
$ 30 389 (26) 238 (24) certified surgeons (P , 0.001), to be treated by male
Male No. (%) 1349 (89) 922 (87)
Board-certified No. (%) 1287 (85) 959 (90)
surgeons (P , 0.001), and to be treated at smaller
Hospital characteristics hospitals (P , 0.001), at non-teaching hospitals (P ,
Bed size No. (%) 0.001), and at hospitals with radiation facilities (P ,
, 100 61 (4) 105 (10) 0.001).
100–249 312 (21) 144 (14)
250–499 565 (37) 471 (44)
$ 500 576 (38) 341 (32)
Teaching hospital No. (%) 974 (64) 61 (6) Standards of Care
Radiation facility No. (%) 1060 (70) 714 (67) The rate and confidence interval for each standard of
Cancer program No. (%) 1384 (91) 584 (55) care are presented in Table 2. The rates for axillary
SD: standard deviation. lymph node dissection, radiation therapy after breast-
a
The denominator used to calculate percentages for some variables is less than the total sample size conserving surgery, and chemotherapy among pre-
due to missing values. menopausal women with positive lymph nodes
b

c
Defined as the median household income for the patient’s zip code.
were . 80% in both states (Table 2). Only the rate for
Defined as the percentage of people in the patient’s zip code with a high school education.
axillary lymph node dissection differed significantly
306 CANCER July 15, 1998 / Volume 83 / Number 2
TABLE 2
Quality Indicator by State
State
Massachusetts
Quality/Indicator % (95% CI)
Radiation therapy after breast-
conserving surgerya 84 (81, 86)
Axillary lymph node dissectionb 81 (79, 83)
Chemotherapyc 97 (94, 100)
Hormonal therapyd 63 (50, 77)
95% CI: 95% confidence interval.
a
Rate for women who underwent breast-conserving surgery (N 5 888 [Massachusetts] and N 5 404
[Minnesota]).
b
Rate for women with known axillary lymph node status (N 5 1498 [Massachusetts] and N 5 1061
[Minnesota]).
c
Rate for premenopausal women with positive lymph nodes (N 5 133 [Massachusetts] and N 5 95
[Minnesota]).
d
Rate for postmenopausal women with positive lymph nodes and positive estrogen receptor status
(N 5 52 [Massachusetts] and N 5 116 [Minnesota]).
between states; more women in Minnesota underwent
the procedure than in Massachusetts.
The rate for hormonal therapy among postmeno-
pausal women with positive lymph nodes and positive
estrogen receptor status was not as high as the rates
for the other patterns of care (Table 2). However, the
number of women eligible for therapy was small (N 5
52 and N 5 116 in Massachusetts and Minnesota,
respectively). For example, in Minnesota the denom-
inator for this rate was low because we excluded from
among the women who were postmenopausal (N 5
784), those with negative lymph nodes (N 5 506),
those who did not undergo axillary lymph node dis-
section (N 5 101), those with positive lymph nodes
who did not undergo estrogen receptor testing (N 5
16), and those with positive lymph nodes who did
undergo estrogen receptor testing but whose result
was unknown or borderline or whose specimen was
too small for analysis (N 5 16) or whose result was
negative (N 5 29). The same pattern existed for Mas-
sachusetts. Nevertheless, among those who were in-
cluded in the denominator, this rate may be low be-
cause some women could have been treated with
chemotherapy followed by hormonal therapy (as sug-
gested by work published since the most recent guide-
lines6,7). Because we collected our data from patients
an average of 3–5 months after diagnosis, patients may
not have begun hormonal therapy when we attempted
to document treatment. Therefore, in the absence of
long term follow-up, it might be best to assess the
proportion of postmenopausal women with positive
lymph nodes who were treated with chemotherapy
and/or hormonal therapy. The proportion of these
women who received any drug therapy (hormonal
therapy and/or chemotherapy) was 92% (95% CI, 87,
96) in Massachusetts and 90% (95% CI, 86 –95) in
Minnesota.
Minnesota
Correlates of Care
% (95% CI)
Radiation Therapy after Breast-Conserving Surgery
86 (83, 90) In Massachusetts, older women were less likely than
94 (92, 95) younger women to receive radiation therapy after
94 (89, 100) breast-conserving surgery (Table 3). In Minnesota, the
59 (51, 68)
oldest women (age $ 80 years) were less likely than
younger women to receive radiation therapy.
In Massachusetts, patients of surgeons who had
been practicing longer and patients treated in hospi-
tals with . 100 beds were more likely than other
women to receive radiation therapy after breast-con-
serving surgery. Neither hospital nor surgeon charac-
teristics were related to the receipt of radiation ther-
apy in Minnesota; however, women who lived in areas
with a median income , $40,000 were more likely
than other women to receive treatment.
Axillary Lymph Node Dissection
In both states, the likelihood of undergoing axillary
lymph node dissection was lower for older patients
than for younger patients (Table 3). In addition, in
Massachusetts the likelihood of undergoing the pro-
cedure was less for those treated in teaching hospitals
than for those treated in nonteaching hospitals and
was greater for those treated hospitals with $ 500 beds
than for those treated in smaller hospitals (Table 3).
DISCUSSION
The rates for three of the four standards of care stud-
ied were . 80% in both Massachusetts and Minnesota.
Only the rate for hormonal therapy among postmeno-
pausal women with positive lymph nodes and positive
estrogen receptor status was low. This rate may ap-
pear low because we followed women for , 6 months,
on average, after primary therapy. It is possible that
some of these women initially treated with chemo-
therapy ultimately may be treated with hormonal
therapy. When we considered treatment with any
form of drug therapy as our quality indicator for post-
menopausal women with positive lymph nodes, we
found that . 90% of these women in either state met
this criterion. Overall, the level of compliance with
quality indicators is encouraging, especially relative to
data collected in 1988 for radiation therapy after
breast-conserving surgery and for adjuvant therapy.16
Unlike choice of surgery, in which the optimal rates
for breast-conserving surgery and mastectomy are not
known, the rates for radiation therapy after breast-con-
 Breast Carcinoma Quality of Care/Guadagnoli et al. 307

TABLE 3
Adjusteda Odds Ratios (and 95% Confidence Intervals) for Radiation Therapy after Breast-Conserving Surgery
and Axillary Lymph Node Dissection in Massachusetts and Minnesota

Quality indicator

Radiation therapy after breast-

conserving surgery Axillary lymph node dissection

Characteristic Massachusetts Minnesota Masachusetts Minnesota

Age (yrs)
, 50 -- -- -- --
50–59 0.4 (0.2, 0.9) 4.2 (0.8, 23.4) 0.8 (0.4, 1.5) 5.3 (0.6, 46.6)
60–69 0.5 (0.2, 1.1) 2.6 (0.6, 12.7) 0.3 (0.2, 0.5) 0.9 (0.3, 3.3)
70–79 0.3 (0.1, 0.8) 0.3 (0.1, 1.3) 0.1 (0.1, 0.2) 0.2 (0.1, 0.7)
$ 80 0.05 (0.02, 0.1) 0.03 (0.01, 0.2) 0.03 (0.02, 0.1) 0.1 (0.02, 0.3)
Educationb (% high school,
zip code)
, 70 -- -- -- --
70–79 1.0 (0.5, 2.3) 1.4 (0.1, 37.7) 0.7 (0.4, 1.2) 0.6 (0.1, 5.8)
80–89 1.3 (0.5, 3.3) 0.4 (0.01, 10.2) 0.8 (0.4, 1.6) 0.6 (0.1, 5.2)
$ 90 1.5 (0.5, 4.1) 1.6 (0.1, 52.6) 0.8 (0.4, 1.6) 0.8 (0.1, 9.0)
Median household incomec
, $40,000 1.6 (0.9, 3.1) 3.7 (1.1, 11.9) 1.4 (0.9, 2.2) 1.4 (0.6, 3.4)
HMO member 1.1 (0.6, 2.0) 2.7 (0.9, 8.3) 0.7 (0.5, 1.1) 1.3 (0.6, 2.9)
Urban residence 0.4 (0.1, 1.8) 6.4 (0.8, 52.2) 0.7 (0.3, 1.6) 1.3 (0.4, 4.6)
Male surgeon 1.4 (0.5, 3.5) 1.0 (0.2, 5.6) 2.0 (0.9, 4.5) 1.8 (0.5, 6.8)
Surgeon Board-certified 0.8 (0.4, 1.6) 0.8 (0.1, 6.2) 1.3 (0.7, 2.3) 0.4 (0.1, 1.9)
Years since graduation
, 10 -- -- -- --
10–19 3.2 (1.1, 9.4) 0.9 (0.1, 7.3) 1.1 (0.4, 2.7) 1.1 (0.3, 4.4)
20–29 7.3 (2.2, 23.9) 1.2 (0.2, 9.3) 1.1 (0.4, 2.8) 0.9 (0.2, 3.9)
$ 30 3.8 (1.2, 12.0) 1.3 (0.2, 10.7) 1.3 (0.5, 3.2) 1.9 (0.4, 8.6)
Teaching hospital 1.0 (0.3, 3.4) 3.7 (0.3, 45.3) 0.2 (0.1, 0.5) 0.5 (0.1, 1.6)
Bed size
, 100 -- -- -- --
100–249 9.1 (2.4, 35.1) 0.3 (0.03, 2.1) 2.1 (0.9, 4.7) 1.7 (0.4, 6.8)
250–499 6.5 (1.2, 35.9) 0.3 (0.02, 3.7) 2.0 (0.8, 5.5) 2.5 (0.5, 13.6)
$ 500 7.9 (1.3, 50.5) 0.1 (0.01, 1.7) 5.8 (1.9, 17.6) 3.3 (0.6, 20.0)
Cancer program 0.9 (0.3, 3.3) 0.9 (0.3, 3.2) 0.8 (0.4, 1.6) 0.8 (0.3, 1.8)
Radiation facility 0.5 (0.1, 2.2) 2.8 (0.7, 10.4) 2.5 (0.9, 6.2) 0.6 (0.2, 1.7)

HMO: health maintenance organization.

a
Other covariates included in the models: marital status, severity of comorbid disease, past history of breast carcinoma, past history of other cancer, stage of disease
(not included in the model for axillary lymph node dissection), and year of diagnosis (Massachusetts models only).
b
Defined as the percentage of people in patient’s zip code with a high school education.
c
Defined as the median household income for patient’s zip code.

serving surgery, axillary lymph node dissection, and ad-
juvant therapy are expected to be as close as possible to
100%. However, it is unlikely that rates of exactly 100%
will ever be attained. Measurement error always will
exist with respect to determining who is eligible for treat-
ment and who received treatment. However, the pri-
mary reason for , 100% adherence may be because our
knowledge of what constitutes good care for early stage
breast carcinoma is evolving continuously. For example,
it has been suggested that radiation therapy can be
avoided in some patients. Radiation therapy after breast-
conserving surgery has been shown to significantly de-
crease the risk of breast carcinoma recurrence, but does
not appear to influence survival.23–26 Selected patients
may be at very low risk for breast carcinoma recurrence
without radiation. Patients age . 55 years treated with
extensive breast-conserving surgery such as quadrante-
ctomy, or patients age . 70 years with small tumors
treated with tamoxifen, appear to have low risks for
recurrence after surgery alone.27,28 Additional informa-
tion will be forthcoming from ongoing randomized trials
evaluating the role of radiation therapy after breast-con-
serving surgery in patients age . 65 years treated with
tamoxifen.
308 CANCER July 15, 1998 / Volume 83 / Number 2
Similarly, the necessity of performing axillary
lymph node dissection in all patients with early stage
disease has been called into question.29,30 Specifically,
patients with a low frequency of axillary lymph node
metastases may be identified based on tumor size,
histology, or other factors.31–33 More important, the
decision to use systemic adjuvant therapy often is not
affected by the results of staging of the axilla, and in
patients who were determined to be lymph node neg-
ative based solely on clinical examination, axillary ra-
diation is associated with low recurrence rates and
offers an effective alternative to surgical treatment of
the axilla.30,34,35 Increased awareness of these results
may have contributed to the lack of surgical treatment
of the axilla in 19% and 6% of women in Massachu-
setts and Minnesota, respectively.
Although this may suggest why , 100% compli-
ance occurred, it does not explain the difference in
rates of axillary lymph node dissection between states.
This difference likely is due to differences in surgical
practice between states. The rate of mastectomy is
considerably higher in Minnesota than in Massachu-
setts9; therefore, axillary lymph node dissection has a
higher likelihood of occurring in Minnesota.
It is interesting that in both states advancing age
was associated with a markedly lower likelihood of
axillary lymph node dissection and radiation therapy
after breast-conserving surgery. These results are sim-
ilar to those recently published by Ballard-Barbash et
al.35 Additional studies are required to determine
whether this represents appropriate standards of
treatment or age bias.
This study has several limitations. First, its results
may not generally be applicable to other states. Sec-
ond, the results for Minnesota may not be generaliz-
able to the whole state because patients were not
studied from a random sample of hospitals. However,
the number of hospitals studied was large and they
represented the majority of admissions in the state.
Third, the patient accrual period for the Massachu-
setts and Minnesota samples did not overlap com-
pletely. We did not include data collected after 1993 in
Minnesota because we were conducting an experi-
ment there to change physician practice patterns rel-
ative to the treatment of early stage breast carcinoma.
Data collected after 1993 reflect practice after the in-
tervention took place, and therefore were not compa-
rable to the situation in Massachusetts, in which an
intervention did not take place. Fourth, the response
rates for the patient and physician surveys were low in
Massachusetts. This should not create a serious bias
for the standard-of-care rates we present because we
used other sources of data when survey data were
missing and because survey data were not the primary
source of data for all indicators. However, this does
indicate that standard-of-care rates could be based on
different sources of data, and raises the issue of con-
sistency of response across data sources. To address
this issue, we assessed, for each state, whether the
different data sources agreed with each other for pa-
tients with data available from multiple sources. We
found agreement rates to be quite high ($ 92%) for
each of the indicators.
We used a variety of data sources to construct our
standard-of-care rates. Our primary sources of data
(the medical record and surveys of patients and phy-
sicians)are expensive and time-consuming. Those in-
terested in documenting the quality of care provided
to women with early stage breast carcinoma on an
ongoing basis may not have the resources necessary to
use these data sources. The utility of using existing
data such as data collected by tumor registries for
ongoing quality assessment should be explored.
Based on data from Massachusetts and Minnesota
for 1993–1995, physician practice appears to be con-
sistent with results of national consensus conferences
and clinical trials dealing with treatment of early stage
breast carcinoma. For practices demonstrated to be
associated definitively with better prognosis (for ex-
ample, chemotherapy for premenopausal women with
positive lymph nodes) or to be important with respect
to prognosis (axillary lymph node dissection) we ob-
served high rates of compliance. Practice that appears
contrary to established standards actually may reflect
new research published after dissemination of the
guidelines rather than the delivery of inappropriate
care. Given the continuous output of new scientific
data, investigators and policy makers who study the
quality of care will need to establish general methods
for distinguishing between nonadherence to stan-
dards of care and innovative care that supersedes
them.
REFERENCES
1. National Institute of Health. National Institute of Health
Consensus Development Conference Statement. The treat-
ment of primary cancer: management of the local disease.
June 5, 1979.
2. National Institute of Health. National Institutes of Health
Consensus Development Conference Statement. Treatment
of early stage breast cancer, June 18 –21, 1990.
3. NIH Consensus - Development Statement. Adjuvant che-
motherapy of breast cancer. The NIH Consensus Develop-
ment Panel. N Engl J Med 1980;303:831–2.
4. National Institute of Health. Adjuvant chemotherapy for
breast cancer. JAMA 1985;254:3461–3.
5. NIH Consensus Conference. Treatment of early-stage breast
cancer. JAMA 1991;265:391–5.

6. Early Breast Cancer Trialists’ Collaborative Group. System-
atic treatment of early breast cancer by hormonal, cytotoxic,
or immune therapy: 133 randomised trials involving 31,000
recurrences and 24,000 deaths among 75,000 women. Part 2.
Lancet 1992;339:71– 85.
7. Glick JH, Gelber RD, Goldhirsch A, Senn J-H. Meeting high-
lights: adjuvant therapy for primary breast cancer. J Natl
Cancer Inst 1992;84:1479 – 85.
8. Goldhirsch A, Wood WC, Senn H-J, Glick JH, Gelber RD.
Meeting highlights: international consensus panel on the
treatment of primary breast cancer. J Natl Cancer Inst 1995;
87:1441–5.
9. Guadagnoli E, Weeks JC, Shapiro CL, Gurwitz JH, Borbas C,
Soumerai SB. The use of breast-conserving surgery for treat-
ment of early stage breast cancer. J Clin Oncol 1998;16:
101– 6.
10. Lazovich D, White E, Thomas DB, Moe RE. Underutilization
of breast-conserving surgery and radiation therapy among
women with stage I or II breast cancer. JAMA 1991;266:
3433– 8.
11. Nattinger AB, Gottlieb MS, Veum J, Yahnke D, Goodwin JS.
Geographic variation in the use of breast-conserving treat-
ment for breast cancer. N Engl J Med 1992;326:1102–7.
12. Nattinger AB, Gottlieb MS, Hoffman RG, Walker AP, Good-
win JS. Minimal increase in use of breast-conserving surgery
from 1986 to 1990. Med Care 1996;34:479 – 89.
13. Young WW, Marks SM, Kohler SA, Hsu AY. Dissemination of
clinical results: mastectomy versus lumpectomy and radia-
tion therapy. Med Care 1996;34:1003–17.
14. Farrow DC, Hunt WC, Samet JM. Geographic variation in
the treatment of localized breast cancer. N Engl J Med 1992;
326:1097–101.
15. Lee-Feldstein A, Anton-Culver H, Feldstein PJ. Treatment
differences and other prognostic factors related to breast
cancer survival. JAMA 1994;271:1163– 8.
16. Hand R, Sener S, Imperato J, Chmiel JS, Sylvester J, Fremgen
A. Hospital variables associated with quality of care for
breast cancer patients. JAMA 1991;266:3429 –32.
17. Gornick ME, Eggers PW, Reilly TW, Mentnech RM, Fitter-
man LK, Kucken LE, et al. Effects of race and income on
mortality and use of services among Medicare beneficiaries.
N Engl J Med 1996;335:791–9.
18. Krieger N. Overcoming the absence of socioeconomic data
in medical records: validation and application of a census-
based methodology. Am J Public Health 1992;92:703–10.
19. American Hospital Association. The AHA guide to the health
care field. Chicago: American Hospital Association, 1993.
20. Folio Associates, Inc. Folio’s Medical Directory of Massa-
chusetts. Boston: Folio Associates, Inc., 1993.
21. Greenfield S, Blanco DM, Elashoff RM, Ganz PA. Patterns of
Breast Carcinoma Quality of Care/Guadagnoli et al. 309
care related to age of breast cancer patients. JAMA 1987;257:
2766 –70.
22. Hosmer DW, Lemeshow S. Applied logistic regression. To-
ronto: John Wiley & Sons, 1989.
23. Clark R, McCulloch P, Levine M, Lipa M, Wilkinson RH,
Mahoney LJ, et al. Randomized clinical trial to assess the
effectiveness of breast irradiation following lumpectomy
and axillary dissection for node-negative breast cancer.
J Natl Cancer Inst 1992;84:683–9.
24. Veronesi U, Luini A, Del Vecchio M, Greco M, Galimberti V,
Merson M, et al. Radiotherapy after breast-preserving sur-
gery in women with localized cancer of the breast. N Engl
J Med 1993;328:1587–91.
25. Liljegren G, Holmberg L, Adami H-O, Westman G, Graffman
S, Bergh J. Sector resection with and without postoperative
radiotherapy for stage I breast cancer: a randomized trial.
J Natl Cancer Inst 1994;86:717–22.
26. Fisher B, Redmond C. Lumpectomy for breast cancer: an
update of the NSABP experience. Monogr Natl Cancer Inst
1992;11:7–13.
27. Nemoto T, Patel J, Rosner D, Daco TL, Schuh M, Penetrante
R. Factors affecting recurrence in lumpectomy without irra-
diation for breast cancer. Cancer 1991;67:2079 – 82.
28. Cady B, Stone M, Wayne J. New therapeutic possibilities in
primary invasive breast cancer. Ann Surg 1993;183:338 – 47.
29. Lin P, Allison D, Wainstuck J, Miller KD, Dooley WC, Fried-
man N, et al. Impact of axillary lymph node dissection on
the therapy of breast cancer patients. J Clin Oncol 1993;11:
1536 – 44.
30. Silverstein M, Gierson E, Waisman J, Gierson ED, Colburn
WJ, Senofsky GM. Axillary lymph node dissection for T1a
breast carcinoma: is it indicated? Cancer 1994;73:664 –7.
31. Chadha M, Chabon A, Friedmann P, Vikram B. Predictors of
axillary lymph node metastases in patients with T1 breast
cancer. A multivariate analysis. Cancer 1994;73:350 –3.
32. Ravdin P, DeLaurentis M, Wenger C, Clark GM. Can prog-
nostic factors be used to predict the nodal status of breast
cancer patients? Proc Am Soc Clin Oncol 1994;13:56.
33. Recht A, Pierce S, Abner A, Vicini F, Osteen RT, Lover SM, et
al. Regional node failure after conservative surgery and ra-
diotherapy for early stage breast carcinoma. J Clin Oncol
1991;9:988 –96.
34. Haffty B, McKhann C, Beinfield M, Fischer D, Fischer JJ.
Breast conservation therapy without axillary node dissec-
tion. A rational treatment strategy in selected patients. Arch
Surg 1993;128:1315–9.
35. 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;88:716 –26.