Cost Effectiveness and Other Assessments of Adjuvant Therapies for Early Breast Cancer

Cost Effectiveness and Other Assessments of Adjuvant Therapies for Early Breast Cancer

ABSTRACT: The 1992 metaanalysis of adjuvant therapies after surgery in early breast cancer summarizes the most extensively studied of all cancer treatments via randomized controlled trials. This study found overall benefits with use of adjuvant therapies, and their expanded use outside the clinical trial setting was assumed to be effective and implied to be cost effective. Thus, the primary remaining questions are which form of adjuvant therapy to use and how to identify which patients are unlikely to benefit. In British Columbia, the effectiveness of adjuvant therapy outside the clinical trial setting was reassuringly similar to the metaanalysis efficacy. Our decision analysis model of hypothetical cohorts of women with early breast cancer confirmed that the efficacy of adjuvant treatment is the primary determinate of its incremental cost effectiveness. Future cost-effectiveness and quality of life assessments should move from hypothetical cohorts assessed via models to prospective data collected within clinical trials or integrated health delivery system. [ONCOLOGY 9(Suppl):129-134, 1995]


Economic and quality of life outcomes are major concerns in the
management of early breast cancer. The potential clinical categories
of early breast cancer amenable to economic, quality of life,
and the general broad category of outcome assessments include
screening primarily using mammography, evaluation of suspected
lesions, primary surgical management, staging, adjuvant therapies,
and monitoring. In this commentary, we restrict our discussion
to the use of adjuvant therapies.

There is no shortage of data or recommendations related to the
adjuvant treatment of early breast cancer. Except for studies
in interventional cardiology, there is no area of medicine, and
no other area in oncology, that has as rich a source of data to
guide clinical treatments. The 1992 metaanalysis of all randomized
trials initiated from 1948 to 1985 and including more than 75,000
patients is a landmark event [1]. It showed reductions in relative
risk with use of adjuvant therapy of up to 30%. The expanded use
of adjuvant therapy outside the clinical trial setting was assumed
or implied to be cost effective. In addition, the National Surgical
Adjuvant Breast and Bowel Project (NSABP) has conducted multiple
clinical trials addressing the primary surgical management of
early breast cancer as well as adjuvant treatments.

Guidelines for treatment exist for all strata of nodal and hormone-receptor
status. The most widely recognized recommendations are from the
1990 NIH Consensus Conference and the 1992 and 1995 St. Gallen
conferences [2,3]. The general conclusion of these conferences
is that almost all patients with clinical and pathologic stage
I or II breast cancer, regardless of age or menopausal, nodal,
or receptor status, should be offered some form of adjuvant therapy
in hopes of improving disease-free and overall survival. The
primary remaining questions are not whether to use adjuvant therapy,
but which form to use and how to identify patients who are unlikely
to experience a meaningful increase in these outcomes.

In this commentary, we wish to discuss the difference between
efficacy and effectiveness of a treatment, provide some information
related to the discordance between recommendations and guidelines
and the actual use of adjuvant therapies in clinical care, review
some highlights of the cost-effectiveness assessments that our
team has done on adjuvant therapies, and make recommendations
on the future design of economic assessments in clinical trials
and general practice. The reader is encouraged to see other articles
in this supplement (Smith and Hillner; Weeks) for further discussion
on decision analysis and quality of life studies.

Efficacy vs Effectiveness

Efficacy refers to a treatment that does more good than harm among
those who receive it. Efficacy is established in the controlled
setting of a clinical trial. This reflects, as near as possible,
ideal conditions with selected patients.

In contrast, a treatment's effectiveness is determined by whether
it does more good than harm in those to whom it is offered. This
specifically refers to general populations that may have less
selected patients with more variations in comorbidity, clinician
involvement, patient compliance, follow-up costs, etc. Effectiveness
refers to the generalizability of an intervention, while efficacy
refers to internal validity.

As discussed earlier, the findings of the efficacy of adjuvant
treatments reported in the metaanalysis were obtained from randomized
clinical trials. However, previous population-based studies from
the late 1980s have either not found a benefit in survival or
could not separate a benefit due to lead-time bias (increased
survival due to early detection). Therefore, until recently, there
has been some uncertainty about whether adjuvant therapy in general
populations leads to benefits equal to those found in clinical

For this reason, the study by Olivotto and colleagues from British
Columbia is particularly important [4]. These investigators took
advantage of how cancer care is recommended and delivered in their
province to address the impact of adjuvant therapy for breast
cancer. In British Columbia, there is a single centralized cancer
agency that keeps a population-based registry. In addition, an
ongoing consensus process makes province-wide recommendations
for cancer treatments. Since medical care is an entitlement program,
ability to pay is not a barrier to the use of adjuvant therapy
as may occur in the United States. During the years studied, the
use of screening mammography was not common, and the incidence
of new breast cancer cases was stable. Therefore, the chance of
a substantial lead-time bias is unlikely.

As shown in Table 1, the investigators studied three cohorts of
patients and assessed their disease-free and overall survival
before and after the adoption of adjuvant therapy recommendations
for these groups. The cohorts were followed for 7 to 16 years
with less than 2% being lost to follow-up. The findings of a 32%
reduction in the chance of dying of breast cancer in premenopausal
and 20% in postmenopausal women between 1974 and 1984 is similar
to the findings from the metaanalysis. Therefore, if widely used,
these therapies do produce benefits similar to those found in
randomized trials.

We have explored the use of adjuvant therapies in a different
population. Our study links the clinical data reported to the
Virginia Cancer Registry with Medicare claims in elderly Virginia
women [5]. We have currently completed analyses related to the
use of adjuvant therapies from 1985 to 1989. In stage II (node-positive)
breast cancer, the use of adjuvant therapies in these years was
substantially lower than that observed in British Columbia. We
found that about 45% of all women with node-positive cancer received
an adjuvant therapy (see Figure 1). In addition, we observed a
substantial age variation in the use of adjuvant therapies, with
hormonal therapy being constant at 30% to 35% in all age groups,
while chemotherapy declined from about 20% in the 65- to 69-year-old
group to less than 3% in those over age 80.

This analysis was limited by the lack of estrogen-receptor data.
Other workers have found that in about 80% of elderly women, breast
cancer is hormone-receptor positive. However, in our study, only
9% of node-negative and 33% of node-positive elderly women received
hormonal therapy.

Estimating Cost Effectiveness of Adjuvant
Therapies Using Decision Analysis

Metaanalysis, decision analysis, and cost-effectiveness analysis
are conceptually related quantitative methods used to combine
information to arrive at a summary conclusion [6]. The historical
impetus for the development of each of these three methods grew
out of the need to resolve uncertainty: for metaanalysis, uncertainty
about conflicting results in the medical literature; for decision
analysis, uncertainty about management of clinical problems; and
for cost-effectiveness analysis, uncertainty about how to best
allocate finite resources.

Although each of these methods appears superficially simple, their
proper conduct can be complex and may rely on multifaceted methodology.
The skills required to perform these type of analyses include
extensive knowledge of research study design, practical skills
and knowledge of the limitations and data collection,
statistics, and critical appraisal of the published literature.
These skills are also used in other areas of outcomes research
and are the cornerstone of practice guidelines development.

Decision analysis is a quantitative approach that assesses the
relative value of different decisions or decision options. In
a medical setting, these typically are physician actions such
as testing or treatments [7]. The information from a decision
analysis is derived from a hypothetical cohort of patients and
is used to recommend management for an individual patient or to
formulate policy recommendations for a group of similar patients.
As outlined in an earlier article in the symposium (Smith and
Hillner), decision analysis begins by systematically breaking
a problem down into its component parts and creating a decision
tree to represent the components and decision options. The uncertainties
in the various components (trees) are identified. The medical
literature or expert opinion is used as a source to estimate probabilities
and define the range of uncertainties around these probabilities.
The values for each outcome, such as survival or quality-adjusted
survival, are measured or inferred.

Node-Negative Disease and Other Questions-The previously
described metaanalysis was originally reported in 1988 [11] and
subsequently updated in 1992 [1]. In 1988, the National Cancer
Institute issued a Clinical Alert to all physicians recommending
that all node-negative breast cancer patients be considered for
treatment with adjuvant chemotherapy [12]. The question of whether
to recommend chemotherapy for these patients has focused on the
difference in the relative and absolute benefits of chemotherapy,
since women with node-negative disease have a relatively low risk
of subsequent systemic recurrence. All women with breast cancer
experience anxiety about their potential future survival, but
if adjuvant chemotherapy is given to all node-negative patients,
all would experience the adverse effects of chemotherapy and incur
the costs of treatment, while the majority might not experience
an increase in survival.

For this and subsequent questions, we developed a decision analysis
model based on the conceptual framework outlined in Table 2 [8-10].
The model addresses the question from a societal perspective using
hypothetical cohorts of women who did or did not undergo chemotherapy
and who are subsequently followed using a mathematical Markov
process to assess their lifetime survival and risk of recurrence.
The model describes nine different health states possible after
diagnosis, from wellness to death.

The initial study of node-negative patients used a relative risk
reduction of 30%, based on reports from randomized clinical trials,
and this was subsequently confirmed by the 1992 metaanalysis.
The model considered only direct health care costs based on a
retrospective cost analysis of individuals at one academic health
center and published cost estimates for treating disease recurrence.
The model was done with and without quality of life adjustments
based on the use of a linear analog scale of surrogates. Since
this was a hypothetical cohort, no specific comorbidities were
considered or patients excluded.

The model showed that the benefit of adjuvant chemotherapy was
highly dependent on the likelihood of disease recurrence, which
varied with patient age and biologic features of the tumor. As
shown in Figure 2, an average 45-year-old woman with an estimated
annual recurrence risk of 4% would have survival benefit of an
increase in quality-adjusted life expectancy of about 5 months.
However, for older women with smaller cancers with an annual recurrence
rate of 1%, the benefit would decrease to less than 1 month. In
contrast, in women with an 8% annual probability of recurrence,
such as those with a large estrogen-receptor negative tumor, the
treatment would increase quality-adjusted survival by approximately
8 months.

The model, by translating the results of the metaanalysis into
units that are more intuitively accessible to patients, physicians,
and policymakers, has been used to tailor recommendations to individual
patients and guide policy development. The relative cost effectiveness
of adjuvant chemotherapy for those women who are at average (4%
to 5%) or high risk of recurrence (6% to 8%) is within commonly
accepted ranges of less than $30,000 per quality-adjusted life
year (QALY).

Cost Comparisons-We have done similar assessments exploring
the use of adjuvant chemotherapy in the elderly and the use of
combination chemotherapy and tamoxifen (Nolvadex) in premenopausal
women. For each of these settings, it was found that the absolute
benefit of therapy dominated the cost-effectiveness determinations.
Given the relatively modest cost of chemotherapy, if it is effective
in a community setting, then it is anticipated to have an acceptable
incremental cost-effectiveness ratio based on comparisons with
other commonly performed medical interventions such as hemodialysis
or treatment of hypertension.

As stressed by others in this symposium, there is no defined threshold
of an acceptable or unacceptable incremental cost-effectiveness
ratio. Hemodialysis is a reference value commonly used by default,
since it is paid for as a societal entitlement. Expanded discussions
of the role of cost effectiveness in the evaluation and adoption
of new technologies are available [13-15].

Limitations of Decision Analysis Models-The primary limitations
of these models were in the precision of the cost data and the
relative crudeness in the quality of life assessments. The monitoring
of the costs of the delivery of adjuvant therapy and the subsequent
management of patients with breast cancer has been relatively
little explored outside older assessments of Medicare-eligible
elderly patients. Large prepaid health plans in the United States
have expressed interest in further assessing this question, but,
to date, their work has been primarily for internal purposes and
has not been published.


Loading comments...

By clicking Accept, you agree to become a member of the UBM Medica Community.