Measuring Real-World Breast Cancer Outcomes

Institution: Stanford University
Investigator(s): Allison  Kurian , M.D. -
Award Cycle: 2010 (Cycle 16) Grant #: 16OB-0149 Award: $1,066,225
Award Type: Translational Research Award
Research Priorities
Innovative Treatments>Gene therapy and other treatments: new frontiers



Initial Award Abstract (2010)

Randomized trials have proven that diagnostic and therapeutic interventions can reduce breast cancer mortality. However, breast cancer care in the “real-world” setting of clinical practice, delivered across a broad range of health care systems and patient populations, deviates greatly from the standards set by clinical trials. Disparities in breast cancer survival persist; interventions are often not used as recommended by evidence-based practice guidelines. There are major areas of controversy at the leading-edge of breast cancer care, for which we lack consensus on how to apply emerging research in the clinical setting. Thus, “real-world” breast cancer care remains an uncharted territory where we lack understanding of what drives specific treatment approaches, and more importantly, how different approaches impact individual patients.

To address critical issues surrounding “real world” issues in breast cancer treatment and care, we will investigate the usage patterns of specific treatments across different California institutions. For this, we will build an internet-based tool that enables patients to report their care preferences, and we will determine how patient preferences shape care. Rather than testing a hypothesis, our proposed work will create, and demonstrate the use of, a ground-breaking breast cancer outcomes research methodology, delivering a uniquely informative report on “real world” breast cancer care.

Specifically, we aim to translate the electronic health records (EHRs) across California institutions into a shareable research tool that generates high-resolution “portraits” of breast cancer care, and identifies specific strategies that yield good and bad outcomes. We will expand a developing database, called OncoShare, through which we link records of breast cancer patients at Stanford University Cancer Center and Palo Alto Medical Foundation (PAMF). First, we will analyze patterns of breast cancer care among a retrospective cohort of 2520 women treated at Stanford and PAMF from 2006-2009, focusing on major treatment controversies at the leading-edge of care. Next, we will develop an electronic patient survey that elicits care preferences and health-related quality of life, with links to individual-level EHR data in OncoShare. Finally, we will enroll a prospective cohort of women starting breast cancer treatment at Stanford and PAMF over the course of one year (target N=320), and administer the patient survey before their first course of therapy. In total, we will evaluate patterns of care matched against the treatment controversies, and identify how patient preferences shape care. We will follow the patient cohorts for cancer recurrence and survival, and identify how specific care patterns serve to optimize these long-term outcomes.

OncoShare extracts de-identified data from the EHRs of all patients receiving breast cancer care at Stanford and PAMF from 2006-2009 (N=2520), including demographics, chemotherapy drug orders, laboratory, radiologic, and pathologic test results. A unique innovation in OncoShare is its ability to “mine” physician’s clinic visit notes through “Natural Language Processing” technology, to obtain details of clinical decision-making. In this study we plan to enhance OncoShare with patient-reported data, by developing a survey instrument in partnership with breast cancer advocates (Breast Cancer Connections in Palo Alto), and administering it to a prospective cohort of women starting therapy at Stanford and PAMF. We will use statistical tests to evaluate patterns of care in the retrospective (N=2520) and prospective (target N=320) cohorts, and to determine the impact of patient-reported information on care.

At the conclusion of this project we will: 1) deliver a detailed analysis of breast cancer care across two major California health care institutions; 2) deliver OncoShare: a highly innovative, widely reproducible methodology for integrating medical records and patients’ own reported experiences into a powerful breast cancer outcomes research tool; 3) transform the kinds of questions that can be asked and answered about breast cancer outcomes across the California and U.S. populations, and 4) adapt EHRs into a tool to measure and optimize real-world breast cancer outcomes.




Progress Report 1 (2011)

Although effective interventions are improving breast cancer survival, we know very little about the last and most important step in their translation from bench to bedside: their impact on the "real world" of breast cancer care, outside of a clinical trial. We do not know what drives the success or failure of an intervention – for example, breast conservation as an alternative to mastectomy - in patient care across the population.

We are developing an innovative resource for the study of "real-world" breast cancer outcomes, built from the electronic health records (EHR) at Stanford University Cancer Institute and Palo Alto Medical Foundation (PAMF). Our objective is to use this novel research tool, called OncoShare, to study patterns and outcomes of breast cancer care across the population. Our specific aims are: 1) to investigate how controversial treatments have been used across these institutions over the last decade; 2) to partner with patients and advocates to develop a survey about patients' treatment preferences and decision-making; and 3) to administer this survey prospectively to patients starting breast cancer treatment, to integrate their responses into OncoShare, and to identify key factors that shape treatment decisions and outcomes.

In Year 1, we made substantial progress with Specific Aim 1, merging data resources including the EHR, billing and medication administration records, physicians' visit notes, and tumor registries at the hospital and state level, to develop an analysis cohort of 11,318 breast cancer patients treated at Stanford and/or PAMF from January 1, 2000 to December 31, 2009. We have analyzed patterns of care over this decade, and found significant differences in mastectomy rates according to cancer stage, treating institution, and other key factors. Ongoing analyses focus on the use of chemotherapy for early-stage breast cancer, and the use of emerging diagnostic tests including magnetic resonance imaging (MRI), positron emission tomography (PET), and tumor genomic profiling. We plan to complete this initial portrait of breast cancer care within Year 2.

For Specific Aim 2, we are working with breast cancer advocates and survivors at Breast Cancer Connections, a community advocacy organization, to conduct focus groups on key aspects of breast cancer treatment such as diagnosis, surgery, systemic therapy, radiation therapy, and survivorship. We are using the results to develop a patient survey, which we will use in Years 2 and 3 for prospective patient data collection (Specific Aim 3).

At the end of Year 1, we are on track with our aims, and have succeeded in developing a larger patient cohort, integrating richer sources of data, than we had originally planned. The initial results of Aims 1 and 2 are intriguing, and will inform our progress on Aims 1-3 in Year 2.




Progress Report 2 (2012)

Emerging interventions are improving breast cancer survival; however, we still know very little about their impact on the "real world" of breast cancer care. We have built a highly innovative data resource, called Oncoshare, using the electronic health records (EHRs) at Stanford University Cancer Institute and Palo Alto Medical Foundation (PAMF). Our specific aims are: 1) to investigate how controversial treatments have been used across academic and community institutions; 2) to partner with patients and advocates to develop a survey about care preferences and decision-making; and 3) to administer this survey prospectively to patients, integrate their responses into Oncoshare, and identify key factors that shape treatment decisions and outcomes.

In Year Two, we made significant progress on Specific Aim 1, merging data sources including the EHRs and state-wide California Cancer Registry, to develop an analytical cohort of 12,115 breast cancer patients treated at Stanford and/or PAMF from 2000-2010. Our Oncoshare data resource now includes granular information on race/ethnicity, socioeconomic variables, tumor characteristics including stage, grade, and marker-defined subtype, treatments including mastectomy (unilateral and bilateral), chemotherapy (with detail on agents such as anthacyclines, taxanes, and trastuzumab), radiation therapy, and novel diagnostics including breast magnetic resonance imaging, positron emission tomography, genetic and genomic tests. We observed significant variability in care across patient sub-groups, and will soon publish these findings.

For Specific Aim 2, we are working with breast cancer advocates and survivors at Breast Cancer Connections, a community advocacy organization, to conduct focus groups on the patient experience across the continuum of breast cancer care. We observed a strong emphasis on the need for more personalized care, with adaptation to individual preferences regarding communication style, information needs, and decision-making. We used these focus group results to develop a survey on patients’ preferences and goals in breast cancer care, which we will administer prospectively in Year Three (Specific Aim 3). We anticipate that incorporating patient-reported data into Oncoshare will help to illuminate the variability that we observe in breast cancer care.

At the close of Year Two, we have nearly completed Aims 1 and 2, with very interesting and informative findings. In Year Three, we will proceed as planned with conducting Aim 3 and reporting results of our work.



A simple heuristic for blindfolded record linkage.
Periodical:Journal of the American Medical Informatics Association
Index Medicus:
Authors: Weber SC, Lowe H, Das A, Ferris T.
Yr: 2012 Vol: 19 Nbr: Abs: Pg:e157-e161

Breast cancer treatment across health care systems: linking electronic medical records and state registry data to enable outcomes research.
Periodical:Cancer
Index Medicus: Cancer
Authors: Kurian AW, Mitani A, Desai M, et al.
Yr: 2013 Vol: Nbr: Abs: Pg: