Evaluating Screening Performance in Practice i

Table of Contents
Foreword . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iii
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
The Breast Cancer Surveillance Consortium: An Overview . . . . . . . . . . . . . . . . 3
Structure of the BCSC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
Collecting Data Within the Context of Routine Clinical Practice . . . . . . . . . . . 9
Opening the Doors to New Understanding: Progress on
a Research Agenda . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
What Characteristics of Women Affect the Performance of Screening
Mammography? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
Do Biological Characteristics of Breast Tumors Determine
Whether They Can Be Detected by Screening Mammography? . . . . . . . . . . . 23
What Characteristics of the Radiology Facility, Radiologist,
or Mammographic Technologist Affect the Performance of Screening
Mammography? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
What Characteristics of Mammography Equipment Affect
the Performance of Screening Mammography? . . . . . . . . . . . . . . . . . . . . . . . . 29
Developing Innovative Statistical Approaches to Analyzing Data . . . . . . . . . . 31
Beyond Research: Other BCSC Accomplishments to Date . . . . . . . . . . . . . . . . 33
Extending Research Use of BCSC Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
Collaborating With the American College of Radiology . . . . . . . . . . . . . . . . . 35
Disseminating Information and Collaborating With Other Groups . . . . . . . . . 36
Enhancing the Career Development of Junior Investigators and Students . . . 43
Meeting the Challenges That Lie Ahead . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
Challenges in Conducting Research . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
Challenges in Using BCSC Data to Influence Clinical Practice . . . . . . . . . . . 47
Appendix A: Metrics for Programmatic Evaluation . . . . . . . . . . . . . . . . . . . . . 49
Appendix B: BCSC Publications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
Collaborative Publications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
Publications From Individual Sites . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
Appendix C: Data Forms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
Patient Information Form . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
Radiologist/Technologist Evaluation—Short Form . . . . . . . . . . . . . . . . . . . . . 73
Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
ii The Breast Cancer Surveillance Consortium
 Evaluating Screening Performance in Practice iii

Foreword

We have good news to report about breast cancer early detection. Research has
shown that early detection, combined with effective treatment, can reduce mortality
from this second leading cause of cancer deaths in women. Since the early 1990s,
breast cancer mortality rates have dropped steadily, in large measure due to improve­
ments in screening and treatment. American women have taken these findings to
heart—in 1987, less than 30% of women 40 years old and older had had a mammo­
gram, the primary mode of breast cancer screening. Ten years later, that percentage
had doubled to 67% of women in the same age group, and is now at 70%. American
women have increasingly come to include breast cancer screening as part of their
regular health care.'

Our growing understanding of the value of breast cancer screening and the wide­
spread use of mammography has led to a need to understand this technology as it is
actually practiced in the community. How accurate is screening mammography in
detecting cancer under a variety of conditions? Do differences in the practice of
screening mammography and resulting diagnostic evaluation influence detection
rates, stage at diagnosis, and survival? How can data from research be used to influ­
ence clinical practice? These questions and more are explored by the National
Cancer Institute’s (NCI) Breast Cancer Surveillance Consortium.

A centerpiece of NCI’s goal of eliminating suffering and death due to cancer is the
“discovery-development-delivery” approach to cancer research. Discovery is the
process of generating new information about fundamental cancer processes from the
genetic to the population level. Development is the process of creating and evaluat­
ing tools and interventions that are valuable in detecting, diagnosing, predicting,
treating, and preventing cancer. Delivery involves promoting and facilitating the
application of evidence-based cancer interventions to all people who need them.
Each of these components is integrally related to the others and all three are neces­
sary for future progress. The Breast Cancer Surveillance Consortium, a key program
of NCI’s Division of Cancer Control and Population Sciences, exemplifies the
“delivery” component, and its research portfolio is helping to accelerate the rate at
which proven interventions are put into widespread clinical and public health prac­
tice.
iv The Breast Cancer Surveillance Consortium

I am pleased to introduce this report describing the work of the Breast Cancer
Surveillance Consortium. By linking surveillance data on breast screening practices
with data from population-based cancer registries and by combining the expertise of
seven research sites around the country, the Consortium has been able to address
issues that can be adequately examined only in large samples of women, radiolo­
gists, and mammography facilities drawn from varied geographic and practice set­
tings. The Consortium has made a major scientific contribution by creating a unique
and collaborative research resource and by greatly extending our knowledge about
the factors that influence the accuracy and performance of breast cancer screening
technologies.

Andrew C. von Eschenbach, MD

Director, National Cancer Institute
 Evaluating Screening Performance in Practice 1

Introduction
Detecting cancer early is critically important because, if effective treatment is pro­
vided, the burden of both illness and death can be reduced. Improvements in breast
cancer treatment and early detection have resulted in a steady drop in breast cancer
mortality rates since the early 1990s, but additional efforts are necessary to ensure
that this trend continues.

For decades, breast cancer early detection technologies have centered on x-ray mam­
mography, and it is the only evidence-based screening technology currently avail­
able. A number of scientific and national organizations have published guidelines
supporting periodic breast screening examinations. Other organizations do not make
any specific recommendations but encourage women to discuss the issue with their
health care providers.

Recent studies have caused debates in the scientific community and the media about
the efficacy of screening mammography and the women who are best served by reg­
ular exams. This debate has focused on a number of issues, particularly the age at
which screening should begin, the optimal frequency of screening, the magnitude of
the impact on mortality, and the quality of the data obtained from randomized trials.
These debates have made it all the more important to assess mammography’s per­
formance in clinical practice and clarify its potential for contributing to reduced
breast cancer mortality rates.

The Breast Cancer Surveillance Consortium (BCSC) was established in 1994 to

enhance the understanding of breast cancer screening practices in the United States
and their relation to changes in stage at diagnosis, survival, or breast cancer mortali­
ty. The BCSC is funded and coordinated by the Applied Research Program (ARP) of
NCI’s Division of Cancer Control and Population Sciences (DCCPS). Through inte­
grated programs of genetic, epidemiologic, behavioral, social, applied, and surveil­
lance cancer research, DCCPS examines the causes and distribution of cancer in
populations. It also supports the development and implementation of effective inter­
ventions, and monitors and explains cancer trends in all segments of the population.

The Applied Research Program’s mission is to evaluate patterns and trends in can­
cer-associated health behaviors, practices, genetic susceptibilities, outcomes, and
2 The Breast Cancer Surveillance Consortium

services. Research within ARP is also targeted to identifying, improving, and devel­
oping databases and methods for cancer control-related surveillance, outcomes, and
applied research; maintaining, updating, and disseminating these databases and
methods; and promoting and facilitating their use among investigators. The BCSC’s
activities are carried out as part of ARP’s efforts to monitor and evaluate cancer con­
trol activities in general and in specific populations in the United States and to deter­
mine the influence of these factors on patterns and trends in cancer incidence, mor­
bidity, survival, and mortality. Rachel Ballard-Barbash, MD, MPH, the Associate
Director, Applied Research Program, is the program director for the BCSC.

This report describes the BCSC and its unique research contribution. The first sec­
tion provides an overview of the BCSC’s mission, history, and structure. This
overview is followed by two sections that describe the BCSC’s current areas of
research and other accomplishments to date. Findings from published studies are
described throughout. The report closes with a discussion of the challenges that lie
ahead for the Consortium, both in terms of its research agenda, as well as potential
opportunities for using BCSC data and findings to influence clinical practice.

“The BCSC has proved to be an invaluable resource for all American radiologists, in its col­
lection and dissemination of robust data on the current practice of mammography in a repre­
sentative cross-section of the U.S. All participating radiologists in San Francisco directly ben­
efit by receiving annually a comprehensive set of audit data that are used for continuing quali­
ty improvement. At UCSF, we have used audit data to facilitate the transition to providing
mammography interpretive services only by radiologists who do full-time breast imaging, at a
documented higher level of performance than the usual-care practice of general diagnostic
radiologists. On a personal level, I have used San Francisco Mammography Registry (SFMR)
data in several of my own clinical research studies and collaborated with BCSC investigators
on other studies. I have used SFMR data to facilitate the successful recruitment of breast-
imaging radiologists to UCSF (access to clinical material of this quality and scope almost
guarantees a successful academic career), and to facilitate the successful recruitment of radiol­
ogy residents to one-year fellowships in breast imaging at UCSF (these physicians will be an
important part of the future of mammography in the United States).
I very much look forward to working with the BCSC to develop interactive Internet-based
tools that all American radiologists can use for the same kind of continuing quality improve­
ment that is now available primarily to BCSC participants.”

Edward A. Sickles, MD
Professor in Residence
Department of Radiology
University of California at San Francisco (UCSF) School of Medicine
 Evaluating Screening Performance in Practice 3

The Breast Cancer Surveillance

Consortium: An Overview
Breast cancer is the second leading non-skin cancer among women and the second
leading cause of cancer deaths in women. Although the breast cancer mortality rate
has dropped since the early 1990s, approximately 40,000 women died from the dis­
ease in 2002 and an estimated 211,000 cases have been diagnosed in 2003. Within
these overall numbers, some important disparities persist among various population
groups. For example, although the breast cancer incidence rate is lower for African
Americans than for whites, their mortality rate is higher. Women of other racial and
ethnic groups have incidence and mortality rates that are lower than those of whites
and African Americans.

Source: NCI Progress Review Groups, “Charting the Course: Priorities for Breast Cancer Research—The Report of
the Breast Cancer Progress Review Group.” http://prg.nci.nih/gov/snapshots/Breast-Snapshot.pdf
4 The Breast Cancer Surveillance Consortium

Large randomized clinical trials con­ that, to obtain truly useful informa­
ducted over the last four decades have tion, screening patterns and associated
shown that by detecting breast cancer performance parameters needed to be
at an early stage, mammography, com­ linked to cancer outcomes—stage at
bined with effective treatment, can diagnosis, morbidity, and mortality.
reduce breast cancer mortality, espe­ With these two premises in mind, NCI
cially among women 50 years old and designed the BCSC to:
older. To ensure standardized delivery
of quality mammography services and ➢ Enhance the understanding of
encourage use of this screening tech­ breast cancer screening practices
nology, the Congress passed the in the U.S. through an assessment
Mammography Quality Standards Act of the accuracy, cost, and quality
(MQSA) of 1992. This Act required of screening programs and the
that mammography facilities meet cer­ relation of these practices to
tain quality standards and be certified changes in breast cancer stage at
by an approved accreditation body. diagnosis, survival, or mortality
The Act also authorized the Secretary
➢ Foster collaborative research
of Health and Human Services to
among surveillance consortium
establish a surveillance system that
participants to examine issues such
could provide reliable and comprehen­
as regional and health care system
sive data on the performance of breast
differences in providing screening
cancer screening.
services and subsequent diagnostic
In response to this legislative man­ evaluation
date, the NCI established the Breast
Cancer Surveillance Consortium. Two ➢ Provide a foundation for clinical
premises guided the NCI in designing and basic science research, espe­
this consortium of research sites. The cially basic research on biologic
first was the longstanding recognition mechanisms that can improve
that results from controlled clinical tri­ understanding of the natural histo­
als can differ from the results of ry of breast cancer.
screening that is practiced in commu­
nity settings. To optimally evaluate The BCSC concept was initially tested
breast cancer screening, it needed to through pilot studies carried out at
be studied within the context of rou­ three locations. In 1994, NCI funded
tine clinical practice. The second was three Consortium research sites
 Evaluating Screening Performance in Practice 5

through a cooperative agreement This has allowed the Consortium to

mechanism and then further expanded analyze data pooled across all sites. In
the number of sites in 1995. This addition, the SCC was designed to
expansion allowed the Consortium establish and evaluate data collection
more latitude to explore issues related and quality control procedures and to
to geography, urban-rural differences, help individual sites analyze data from
and racial and ethnic diversity. In their own sites. In 2000, the coopera­
1995, NCI also funded a Statistical tive agreements for the Consortium
Coordinating Center (SCC) to serve as sites were renewed for an additional
the repository of data from all sites. five years.

BCSC: A Snapshot of Progress

Phases Research Group Interactions

Pilot studies, 1990-2 Conducted at three sites

First RFA release, Sites:
Phase I, 1993
3 centers funded ➢ Agree on goals and data
elements
➢ Set up systems

Second RFA release, Phase II, Sites: ➢ Establish governance

1994 and research priorities
Added 3 centers and created ➢ Establish data standards
SCC as supplement to one and Certificates of ➢ Institute publications com­
center Confidentiality for women mittee and management
and providers system
➢ Begin transition from
paper to electronic systems
➢ Improve data editing
➢ Actively conduct research
➢ Begin planning for pooled
analyses

Renewal Sites: ➢ Refine systems

Phase III, 2000-2004
Independent SCC ➢ Add new data ➢ SCC develops interactive
research Web site for
➢ Make major progress in
BCSC sites
site and pooled research
6 The Breast Cancer Surveillance Consortium

The Consortium currently consists of ➢ New Mexico Mammography

the SCC and seven data collection and Project
research sites. Six sites are defined by
➢ San Francisco Mammography
geographic region; the seventh (Group
Registry
Health Cooperative) is defined by
membership in a health maintenance ➢ Vermont Breast Cancer
organization: Surveillance System.

➢ Carolina Mammography Registry The investigators working across these

➢ Colorado Mammography Project sites are a multidisciplinary team that
includes radiologists, primary care cli­
➢ Group Health Cooperative, Center nicians, pathologists, epidemiologists,
for Health Studies statisticians, physicists, and advocates.
➢ New Hampshire Mammography
Network
 Evaluating Screening Performance in Practice 7

BCSC: Principal Investigators and NCI Staff

Carolina Mammography Registry Department of Radiology

Bonnie C. Yankaskas, PhD University of North Carolina
Chapel Hill, NC

Colorado Mammography Project University of Nevada at Reno

Gary Cutter, PhD Reno, NV
Mark Dignan, PhD Kentucky Prevention Research Center
Lexington, KY

Group Health Cooperative Group Health Cooperative, Center for Health Studies
Stephen Taplin, MD, MPH 1994–2003 Seattle, WA
Diana Buist, PhD 2003–Present

New Hampshire Mammography Network Department of Community and Family Medicine

Patricia Carney, PhD Dartmouth Medical School
Hanover, NH and Lebanon, NH

New Mexico Mammography Project Department of Radiology

Charles Key, MD, PhD 1995–1997 Health Sciences Center
Robert Rosenberg, MD 1997–Present University of New Mexico
Albuquerque, NM

San Francisco Mammography Registry Department of Medicine, Epidemiology

Virginia Ernster, PhD 1995–2000 and Biostatistics
Karla Kerlikowske, MD 2000–Present University of California at San Francisco
San Francisco, CA

Vermont Breast Cancer University of Vermont

Surveillance System Office of Health Promotion Research
Berta Geller, EdD Burlington, VT
Don Weaver, MD

Statistical Coordinating Center Group Health Cooperative, Center for Health Studies
William Barlow, PhD Seattle, WA

National Cancer Institute Applied Research Program

Rachel Ballard-Barbash, MD, MPH Division of Cancer Control and Population Sciences
Robin Yabroff, PhD, MBA Rockville, MD
Kathleen Barry
Stephen Taplin MD, MPH 2003–Present
8 The Breast Cancer Surveillance Consortium

Structure of the BCSC

The Consortium has three co-chairs tors of all the sites, a pathologist co­
(the NCI project director, a site princi­ investigator, and the NCI project
pal investigator, and the SCC principal director), a Publications Committee,
investigator). Additional oversight is and Working Groups. Working Groups
provided by a Steering Committee are formed for specific projects and
(composed of the principal investiga­ disband when no longer needed.

Breast Cancer Surveillance Consortium

Carolina
Mammography
Center
Colorado Group

Pu mmit
e

Co
Co ering

Mammography Health

blic tee
itte

Project Cooperative
mm

atio
Ste

ns
Statistical
New Hampshire Coordinating New Mexico
Mammography Center Mammography
Network

San Francisco Vermont Breast

Mammography Cancer Surveillance
Registry System

Working
Groups
 Evaluating Screening Performance in Practice 9

Collecting Data Within the Context of Routine Clinical Practice

Unlike a multicenter clinical trial, practices, such as pathology laborato­
which uses a common protocol and ries, surgical practices that perform
common data collection instruments, breast biopsies, and other medical
the BCSC sites conduct research with­ practices where mammography is per­
in existing health systems and within formed (e.g., obstetrics and gynecolo­
the context of routine clinical practice. gy, internal medicine, and family med­
icine practices).
The BCSC: A Unique Resource
Each participating facility collects
As of October 2003, the Consortium had several distinct types of data about
collected data for more than 1.7 million
women and more than 5 million mammo­
women and their mammographic
grams. Within this group, about 38,000 exams. The data collected about
breast cancers have been detected. women include basic information
The size of the BCSC database, the
longitudinal nature of these data, and the
multidisciplinary teams of participating “I feel lucky to be a part of the BCSC
investigators make the BCSC a unique group that has been working together
resource for understanding breast cancer pooling their information and answering
screening practices and outcomes in the questions of value for women all over the
U.S. U.S. My presence makes me feel a part
of the research process, and I think it also
is a reminder to the researchers and doc­
tors that patients are real people who are
Each BCSC site has developed volun­ waiting for answers.”
tary partnerships with mammography
Bambi Schwartz
facilities in its geographic area. In Patient Advocate, San Francisco
some cases, 100% of facilities in the Mammography Registry
area partner with the site. In other
cases, fewer facilities participate.
Participating facilities represent a about their demographics, health his­
wide range of health care settings, tory, screening history, and current
including traditional fee-for-service health status. Information collected
solo and group radiology practices; about the exam includes the indication
managed care organizations; mobile for the exam, breast density, exam
mammography vans; freestanding assessment, and follow-up recommen­
mammography programs; hospital- dation. As part of ancillary studies,
based services; and nonradiology some sites also collect data about
10 The Breast Cancer Surveillance Consortium

radiologists, such as their specialty, Surveillance, Epidemiology, and End

practice patterns, and perceptions Results (SEER) Program. The
about screening and breast cancer risk. Colorado Mammography Project is
All data collection procedures have linked to its statewide pathology reg­
been approved by each site’s istry. The Carolina Mammography
Institutional Review Board (IRB) and Registry, New Hampshire
are compliant with the Health Mammography Network, and Vermont
Insurance Portability and Breast Cancer Surveillance System
Accountability Act (HIPAA). collect benign and malignant breast
pathology reports from laboratories in
A defining characteristic of the BCSC their defined regions and additionally
is that the data it collects from women link to their respective state cancer
and radiologists/facilities are linked to registries.
cancer outcomes data from popula­
tion-based cancer and pathology reg­ One of the Consortium’s first tasks
istries. This linkage occurs at each after it was established was to deter­
site. Three sites—Group Health mine how to organize these various
Cooperative, the New Mexico types of clinical data so that they
Mammography Project, and the San could be used for research. This
Francisco Mammography Registry— required Consortium investigators to
are linked to registries within NCI’s identify the critical data elements

SEER: A Vital Source of Population-Based Cancer Data

The Surveillance, Epidemiology, and End Results (SEER) Program of the National Cancer
Institute is an authoritative source of information on cancer incidence and survival in the
United States. The SEER Program currently collects and publishes data on all types of cancer
from 11 population-based cancer registries and three supplemental registries. Approximately
26% of the U.S. population is covered by the SEER Program. Information on more than 3 mil­
lion in situ and invasive cancer cases is included in the SEER database, and approximately
170,000 new cases are documented each year within the SEER catchment areas. The SEER
Registries routinely collect data on patient demographics, primary tumor site, morphology,
stage at diagnosis, first course of treatment, and follow-up for vital status. SEER is the only
comprehensive source of population-based information in the United States that includes stage
of cancer at the time of diagnosis and survival rates within each stage. The mortality data
reported by SEER are provided by the National Center for Health Statistics.
 Evaluating Screening Performance in Practice 11

necessary for evaluating screening performance and to develop a consensus on a

standard set of core variables, response categories, definitions for analysis, and stan­
dard definitions of screening and diagnostic mammography. Within this common
data structure, the sites agreed to maintain their own data collection procedures,
developed with their participating mammography facilities, cancer registries, and
pathology databases. These procedures have evolved over time as electronic data
collection methods have gradually supplanted paper-based systems.

How Representative are BCSC Data?

Two important goals of the BCSC are that the data collected reflect mammography practice
as it is performed in the community and that the population of women in the BCSC reflect
the distribution of women in the U.S. who undergo mammography.
A comparison of women represented in the BCSC against 2000 Census data shows that
Consortium sites are located in counties that contain slightly more than 5% of the Nation’s
population. As the following table shows, data in the BCSC reflect the national population in
several important respects.

BCSC Counties All other

U.S. Counties
Sociodemographic Median Family Income $55,189 $50,984
Characteristics Percent Unemployed 3.4% 4.1%
Percent With High School Degree 84.5% 80.2%
Percentage of women aged 40+ 22.0% 22.7%
Sociodemographic Percent Hispanic 6.9% 7.3%
Characteristics in Percent Black 8.9% 10.9%
Women Aged 40+
Data Source: 2000 Census
12 The Breast Cancer Surveillance Consortium

Data Collected by BCSC Sites

From Women

Demographic Variables
➢ Unique anonymous identification number
➢ Zip code
➢ Date of birth
➢ Race (white, black, Asian or Pacific Islander, Native American, other); ethnicity (Hispanic)
➢ Education (1-11 years, 12, 13-15 years, 16 years, 16+ completed years of education)

Health History
➢ Age at birth of first child (year)
➢ Age at menarche
➢ First-degree family history of breast cancer (mother, sister, daughter) and age: <50, >50
➢ Personal history of breast cancer (yes, no)
➢ Personal history of breast biopsy, surgery, or radiation (yes, no)
➢ Procedure history per breast (implants, needle biopsy, surgical biopsy, lumpectomy, mastectomy,
radiation therapy, and reconstruction)

Screening History
➢ Ever screened by mammography (yes, no)
➢ Time since last mammogram (within last year, 1-2 years, 3-4 years, 5 or more years)
➢ Time since last clinical breast examination

Current Health
➢ Menopausal status at examination (pre-, peri-, postmenopausal)
➢ Hormone replacement therapy use at time of examination (yes, no) and type (e.g., estrogen,
estrogen/progestin, over-the-counter supplements)
➢ Presence of symptoms in last three months (nipple discharge or lump; right or left breast)
➢ Main reason for current visit (routine screening, routine follow-up, concerns about breast prob­
lems)

From Radiologists and Technicians

Radiologic Site and Interpreting Mammographer Identification (encrypted)

Dates of Current Examination and Comparison Film
Use of Comparison Mammogram at Time of Evaluation (yes, no)
Indication for Examination
➢ Screening (asymptomatic), evaluation of breast problem (symptomatic), additional evaluation of
recent mammogram, short interval follow-up
 Thank You for previewing this eBook
You can read the full version of this eBook in different formats:

 HTML (Free /Available to everyone)

 PDF / TXT (Available to V.I.P. members. Free Standard members can

access up to 5 PDF/TXT eBooks per month each month)

 Epub & Mobipocket (Exclusive to V.I.P. members)

To download this full book, simply select the format you desire below