Joseph Lipscomb, PhD jlipscosph'emory'edu Professor of Health Policy and Management Rollins School o

1
Joseph Lipscomb, PhD( jlipsco_at_sph.emory.edu
)Professor of Health Policy and
ManagementRollins School of Public Health, Emory
UniversityGeorgia Cancer Survivorship
ConferencePeachtree City, Sept 28, 2007
Using (and Augmenting) Georgia Cancer Registry
Data to Understand and Improve Survivor Outcomes
2
Who Is a Cancer Survivor?(Critical for
Defining the Scope and Breadth of Data Required
for Analysis)

• An individual is considered a cancer survivor
  from the time of diagnosis, through the balance
  of his/her life. Family members, friends, and
  caregivers are impacted.and included in this
  definition.
• (Office of Cancer Survivorship, National
  Cancer Institute http//dccps.nci.nih.gov/ocs/)
• The term cancer survivors refers to those people
  who have been diagnosed with cancer and the
  people in their lives who are affected by the
  diagnosis, including family members, friends, and
  caregivers.
• (from A National Action Plan for Cancer
  Survivorship, issued jointly by the Centers for
  Disease Control and Prevention and the Lance
  Armstrong Foundation http//www.cdc.gov/cancer)

3
What Are the Outcomes of Interest

• Quantity of life (survival)
• Quality of life
• - Symptom bother
• - Functional status (e.g., carrying out
  normal activities of daily living)
• - Health-related quality of life (HRQOL)
  an integrated assessment of the impact of
  health-related events on the individuals overall
  level of self-assessed well-being, often taking
  into account physical, social, and psychological
  functioning
• Economic status of the cancer survivor and
  his/her family

4
Improving the Outcomes of Cancer Care
National Cancer Institute Perspective
Identifying Interventions that Improve Outcomes
Strengthening Science Base for Quality of Care
Measuring Outcomes that Matter
Impacting the Delivery of Care
Monitoring Progress and Identifying New Targets
5
Major Survivorship Research and Policy Issues
Requiring Ongoing Empirical Study (as reflected,
for example, in the NCI research agenda)

• Chronic and late effects of cancer and its
  treatment
• Interventions to reduce or halt these effects
• Health lifestyle and behaviors (to reduce the
  risk of cancer recurrence and improve overall
  quality of life and survival)
• Benefit finding and posttraumatic growth
  following the diagnosis and treatment of cancer
• Family impacts
• Health disparities

6
What Kinds of Data Are Required to Study Such
Survivorship Issues in Georgia?

• Specific, targeted questions can be addressed by
  specific data sources (e.g., expected survival
  can be estimated from high-quality cancer
  registries HRQOL can be estimated at a slice in
  time by well-executed survey of individuals
  diagnosed with cancer)
• But to acquire the capability to address a full
  range of cancer survivorship issues, Georgia
  needs a comprehensive cancer data system that
  capitalizes on a range of current information
  sources and develops additional ones.

7
Components of a Comprehensive State-Based Cancer
Data System for Survivorship Analysis

• Strong cancer registry system -- and the Georgia
  Comprehensive Cancer Registry (GCCR) is strong
  indeed
• Access to medical records/charts of cancer
  survivors, during primary treatment and beyond
  the more electronic the data, the better
• Access to insurance claims data (Medicare,
  Medicaid, private indemnity plans, managed care)
• Information on symptoms, HRQOL, economic
  consequences of cancer from the survivors own
  perspective
• Link it all together to develop, for each
  individual diagnosed with cancer, a rich
  longitudinal picture of what it is like to live
  with, through, and beyond cancer

8
Components of a Comprehensive State-Based Cancer
Data System for Survivorship Analysis

• Such a population-based cancer data system has
  been recommended by the Institute of Medicine
  (Enhancing Data Systems to Improve the Quality of
  Cancer Care, 2000 and From Cancer Patient to
  Cancer Survivor Lost in Transition, 2006).
• Likewise, the forthcoming Georgia Comprehensive
  Cancer Control Plan has, as one of its Data
  Metrics Goals, to expand and enhance data
  collection from existing and new sources.
• But can we get there from here? Do we have
  examples of how cancer registry data can be
  augmented to study the quality and outcomes of
  care over time?
• (Yes we do!)

9
Example 1, the CDC-supported study being
conducted by Emory University investigators on...

• Determinants of Patient Dropout from Cancer
  Treatment and Follow-up

10
Background Motivation

• Numerous patterns-of-care studies report that
  cancer patients frequently do not receive
  guideline-concordant care (Institute of Medicine,
  1999).
• Yet, little is known about patients not
  completing prescribed therapy, the factors that
  predict patient dropout, and even whether
  currently available data systems are adequate to
  support such analyses.

11
Primary Study Goals

• (1) Examine frequency of failure to complete
  prescribed treatment for breast, colorectal,
  lung, and prostate cancers within the 1st year of
  diagnosis among residents of Southwest Georgia
  a 33-county region of the state.

Region of SW GA Cancer Coalition
12
Primary Study Goals

• (2) Investigate tumor-specific and
  patient-related, provider-related, and health
  system-related factors associated with premature
  treatment discontinuation in this largely rural
  population.
• (3) Determine the types of data sources and
  information systems required to achieve goals (1)
  and (2) in a population-based cancer treatment
  environment.
• Additional Study Goals
• (4) Identify interventions to reduce dropout
• (5) Propose strategies to disseminate findings
• (6) Recommend whether longer-term follow-on
  study is warranted

13
Discussion and Conclusions I

• In this rural population, 85 of breast cancer
  patients and 67 of colorectal cancer patients
  overall completed adjuvant chemotherapy
• Those cases who stopped prior to finishing
  treatment were much more likely to do so for
  toxicity and other clinical reasons than for
  refusal/lack of support
• Married CRC patients were significantly more
  likely to complete treatment than patients who
  were single, divorced or separated
• Among non-married breast cancer patients, older
  women and whites were more likely to discontinue
  chemotherapy for reasons other than completion

14
Discussion and Conclusions II

• In all analyses undertaken, white/non-white
  differences in chemotherapy receipt and treatment
  completion were either non-significant or else
  suggested higher completion rates for non-whites
• Future analyses will focus on the relationship
  between treatment plan and compliance, using
  qualitative research to investigate provider
  decision making

15
Example 2funded by CDCs National Program of
Cancer Registries (NPCR)

• Breast and Prostate Cancer
• Data Quality and Patterns of
• Care Study

16
Background and Motivation

• In two influential reports, National Cancer
  Policy Board of the IOM (1999, 2000) emphasized
  both inadequacy of our present understanding of
  cancer care quality and important role of NPCR in
  monitoring and improving quality.
• The two most common cancers in Georgia are breast
  and prostate (gt5,000 cases each in 2004).
• Georgia Comprehensive Cancer Registry is one of
  the nations best and can become still better
  for patterns-of-care and quality-of-care
  assessment.

17
Primary Study Goals

• Describe patterns of care for female breast and
  prostate cancer among Georgia residents, based on
  a stratified random sample of 2,000 cases of BC
  and 2,000 cases of PC diagnosed in 2004
• Examine the degree to which cancer care is
  guideline concordant a frequently used
  approach to quality-of-care assessment
• Investigate factors that may influence patterns
  of care and guideline concordance
• -- tumor-related
• -- patient-related (including
  race/ethnic, socio-
• economic, and geographic
  variables)
• -- provider-related
• -- health system-related
• Assess and improve the quality of the data on
  cancer care and outcomes

18
Examples of Specific Hypotheses (cutting across
both breast and prostate cancer)

• Patients at higher socio-economic status more
  likely to receive guideline-concordant care as
  defined by the National Comprehensive Cancer
  Network (NCCN) guidelines
• Younger patients more likely to receive guideline
  concordant pre-treatment evaluation and care
• Patients comorbidity status will significantly
  influence choice of care, with strength of the
  effect varying by age and stage of disease
• Non-Hispanic whites more likely to receive
  guideline-concordant care than Hispanics and
  non-white race
• Patients seeking pre-treatment evaluation at
  high-volume facilities more likely to receive
  guideline concordant care

19
Organization of the Study

• A 7-state consortium formed by CDC in Oct 2005
  following competitive selection process (RFA
  grant mechanism) GA, CA, LA, KY, MN, NC, and
  WI.
• Critically important element of study in Georgia
  Collaborative relationship between GCCR, GA
  cancer registrars, and Emory investigators.

20
Example 3 Prostate Cancer Outcomes Study
(data from multiple states, NCI-supported)

• OBJECTIVES
• Describe diagnosis and treatment patterns for
  prostate
• cancer among large (N 3,500) diverse group
  of men
• treated in community settings and followed
  periodically
• Assess health-related quality of life among
  prostate cancer survivors at 5 time points over
  5-year period
• Focus on complications associated with prostate
  cancer urinary, bowel, sexual
• Overcome limitations of previous studies,
  including
• -- Patients from single institutions
• -- Quality of life evaluated by doctors, not
  patients
• -- Cross-sectional, not longitudinal

21
Some PCOS Findings

• The prevalence of urinary, bowel, and sexual
  impairments is substantially higher than some
  estimates from referral centers and leading
  academic institutions.
• As part of treatment planning, clinicians should
  consider mens baseline function and age, since
  these also predict subsequent outcomes and
  recovery after treatment.
• PCOS data are increasingly being used as a
  treatment decision resource by patients and
  physicians in the community setting.
• Need for new therapies with lower rates of
  complications.

22
Example 4

• CanCORS
• Cancer Care Outcomes Research and Surveillance
  Consortium A 6-year, 40M, project supported by
  NCI and VA to
• study the impact of targeted interventions on
  patient-centered outcomes
• investigate dissemination of state-of-the-art
  therapies in the community
• examine gaps between best, evidence-based
  clinical practice and actual care in community
• analyze disparities in quality cancer care

23
CanCORS Study Design

• Large observational cohort study of newly
  identified lung and colorectal cancer patients
• -- For lung 5 research teams with N
  4,700
• -- For colorectal 6 research teams
  with N 5,300
• Socio-economically,geographically, and
  race/ethnically diverse samples
• Public-private provider mix large HMOs,
  fee-for-service, VA medical centers

24
CanCORS Study Design

• Rapid case ascertainment lt 3 months after
  diagnosis
• Follow-up patients 12 months after diagnosis
• For each patient, creates a longitudinal profile
  of cancer care by utilization multiple data
  sources
• Investigate structure - process - outcome links
  at the patient, provider, and organizational
  level

25
CanCORS Specific Aims

• 1) To determine how the characteristics
  beliefs of cancer patients and providers and the
  characteristics of health-care organizations
  influence treatments and outcomes,
• spanning continuum of cancer care from diagnosis
  to recovery or death
• 2) To evaluate effects of select processes of
  care on patients survival, health-related
  quality of life, and satisfaction with care

26
Some CanCORS High-Priority Questions

• How and why do processes and outcomes of care
  vary by patient age, race, ethnicity, SES?
• Why do high-volume hospitals tend to have lower
  surgical mortality rates?
• How do patients and physicians go about making
  treatment decisions for metastatic cancer?
• Are symptoms (especially pain and depression)
  treated effectively?

27
But these and other specific studies do not
produce, nor benefit from, a strong, enduring,
population-based cancer care data system.

• How do we build such a system in Georgia to
  understand and improve survivor outcomes?

28
What are the Options?

• (1) Rely primarily on registry data
• - Such data crucial foundation for
  many survivorship studies
• - Sufficient to study impact of
  cancer on quantity of life (mortality survival)
• - But does not address
  quality-of-life and economic consequences of
  cancer and its treatment

29
What are the Options?

• (2) Link registry data with insurance claims data
• -- the linkage of the NCI Surveillance,
  Epidemiology, and End Results (SEER) data with
  Medicare claims has resulted in a data base
  (SEER-Medicare) supporting numerous studies of
  patterns of care, quality of care, and economic
  outcomes for cancer survivors across the U.S.
• -- And, starting with the observation
  period 1999-2002, we now have Georgia-Medicare
• -- Next, we need Georgia-Medicaid and
  also the linkage of private insurance claims data
  to Georgia cancer registry data
• -- Aim create a (near) population-represent
  ative data base linking (most) every Georgia
  cancer registry record with the individuals
  insurance claims files
• At least one significant problem the
  uninsured!

30
What are the Options?

• (3) Extend the follow-up period of existing
  patterns-of-care/quality-of-care (POC/QOC)
  studies
• For example, continue to follow the
  survivors now enrolled in the Breast and Prostate
  Cancer POC/QOC study and the SW Georgia cancer
  care drop-out study
• (4) Launch new POC/QOC studies over time, in a
  strategic fashion to (a) encompass a broad array
  of tumor types and (b) include patient-reported
  outcomes such as symptom bother, HRQOL, and
  economic consequences

31
What are the Options?

• (5) To augment POC/QOC studies that do not
  collect patient-reported outcomes, expand and
  intensify population-based surveys that capture
  symptom bother, HRQOL, economic outcomes, and
  other data important for interpreting the
  experiences of the cancer survivor, caregiver,
  and others touched by cancer

32
What are the Options?

• (6) Continue to follow patients enrolled in
  cancer clinical trials well beyond the end of
  the study, and expand the data collected to
  include patient-reported outcomes
• -- As the Georgia Center for Oncology
  Research and Education (GA CORE) works to
  increase the number of clinical trials across the
  state, this option should receive heightened
  attention

33
What are the Options?

• (7) Over the long term, seek to establish a new
  standard of health care data collection at the
  site of care, to include information on the
  patients socioeconomic status (e.g., education
  level), stage of cancer disease (if relevant),
  comorbid conditions, and patient-reported
  outcomes including symptom bother, HRQOL, and
  economic consequences of disease and treatment.
• -- Aim to make such data part of the
  patient record that can then be abstracted in
  POC/QOC studies, thus reducing need for separate
  surveys

34
Challenges

• Affordability covering the cost of an expanded
  Georgia cancer data system
• Sources state govt, federal govt,
  funded research, foundations, private industry
• Cooperation, Coordination, Collaboration
  multiple public and private entities with a
  strong interest in the cancer survivor and also
  unique perspectives, agendas, and fiscal
  administrative constraints
• Patient Privacy and Confidentiality the
  importance (and the considerable cost) of
  obtaining informed consent and ensuring privacy
  of medical records

35
But we should keep our eyes on the Prize.

• A comprehensive Georgia cancer data system that
  enables us to understand and improve healthcare
  outcomes for cancer survivors across the state

36

• The Opportunity
• and
• The Commitment