Transcript ChristinaClarkex

Maximizing the social impact of cancer registry
data project kickoff meeting
October 28, 2009
Christina Clarke, Ph.D
Northern California Cancer Center
Stanford Cancer Center
 Non-profit
cancer research institute in Fremont,
California; 15 PI’s
 Population sciences group of NCI-designated
Stanford Cancer Center
 Mission: “…understanding the causes,
prevention and detection of cancer and
improving the quality of life for individuals living
with cancer”
 Operate the SEER cancer registry for nine
county region in SF/Monterey Bay Areas
 www.nccc.org or cancer.stanford.edu/nccc
 Provide
orientation to SEER and other
population-based cancer registry data
• content, how collected and organized
 Summarize
efforts to disseminate data
• with perspective of working at regional SEER registry
 Discuss
challenges to changing current
dissemination practices
All 50 states now have mandates for reporting newly
diagnosed cancers* to a regional cancer registry
 State registries funded by some combination of NCI,
CDC, state, and private funds

• 1973: NCI funded nine regions as SEER program
• 1992: CDC funded National Program for Cancer Registries
in other states

Represent complete census of cancer patients for
geographic area
• the only population-based and therefore representative
resources for quantifying progress in war on cancer
* Except basal cell and squamous cell carcinomas of the skin
26% of the total US
population
9 regions with coverage
from 1973-2006
All 18 regions with coverage
1992-2006
seer.cancer.gov

Central body sets clear standards for all procedures,
data items to be collected, and coding
• North American Association of Central Cancer Registries
 Input from American College of Surgeons, NCI, CDC, other
professional groups

At hospital:
• Data abstracted directly from medical record by certified
tumor registrars (CTR)

At central registry:
• Cases consolidated from multiple reports
• Myriad, standardized quality control
• Finding unreported cases

SEER standards require ≥99% completeness

Patient characteristics
• Limited to age, sex, race/ethnicity, address at diagnosis)

Tumor characteristics
• Detailed anatomic site, other pathological and clinical features
• Detail regarding extent of disease spread/stage
• First course of treatment (in first 4 months)

Patient survival time
• All patients followed for vital status for life
• Death status and cause of death from routine linkages with
 state vital statistics
 National Death Index
• Probable living status from routine linkages with
 Social security benefits
 National credit agency records
 Routinely produced
• Incidence rates the number of newly diagnosed cases
during a specific time period
• Mortality rates the number of deaths during a specific
time period
 More sophisticated
• Prevalence new and pre-existing cases for people alive on
a certain date
• Survival* the proportion of patients alive at a given time
after cancer diagnosis
• Lifetime risk the probability of developing or dying of
cancer
* Data collected only by SEER and not most state registries

Structured rosters of cancer statistics
• Rates and trends stratified by
 Cancer sites or types
 Subpopulations of interest (age, race, geography)
 Time: most current vs. trends
• Increasingly available as interactive web tools
 Obtain detailed statistics of interest or explore data
• Packaged into “fact sheets” and other standalone
documents

Particular observations from the data
• Proactive results of cancer surveillance research
• Reactive responses to issues of public interest
NCI “Snapshot of..” fact sheets
Cancer Stat Fact Sheets provide a quick overview of frequently-requested cancer statistics
including incidence, mortality, survival, stage, prevalence, and lifetime risk.
The SEER Cancer Statistics Review (CSR), 1975-2006 includes tables showing cancer statistics
by race, sex, age, and year of diagnosis for the major cancer sites and for all cancers combined.
The report is available in both HTML and PDF formats. And, there's a search interface to generate
custom reports.
Fast Stats links to tables, charts, and graphs of cancer statistics for all major cancer sites by age,
sex, race, time period. The statistics include incidence, mortality, survival and stage, prevalence,
and the probability of developing or dying from cancer. A large set of statistics is available in predefined formats.
Cancer Query Systems provide more flexibility and a larger set of cancer statistics than Fast Stats
but require more input from the user. A larger number of cancer sites are available, and you have
the ability to customize the format of the output into tables, graphs, and delimited format.
State Cancer Profiles is a comprehensive system of dynamic maps and graphs enabling the
investigation of cancer trends at the national, state, and county level.
Cancer Mortality Maps and Graphs shows geographic patterns and time trends of cancer death
rates for the time period 1950-1994 for more than 40 cancers.
Cancer Data and Statistics Tools
United States Cancer Statistics
The United States Cancer Statistics: Incidence and Mortality Web-based report contains official federal
government cancer statistics for new cancer cases and deaths.
State Cancer Facts
The State Cancer Facts application provides state-specific data on lung, colorectal, female breast, and
prostate cancers, including the estimated number of new cancer cases, cancer deaths, and the ageadjusted mortality rates for cancer deaths by race for each state.
National Environmental Public Health Tracking Network
This tool unites environmental information from across the United States with information for some
chronic conditions, including cancer. For more information about CDC's Environmental Public Health
Tracking Network, visit CDC's feature page.
Chronic Disease Cost Calculator
This downloadable tool helps states estimate state Medicaid expenditures for congestive heart failure,
heart disease, stroke, hypertension, cancer, and diabetes using customized inputs such as prevalence
rates and treatment costs.
The Global Cancer Atlas Online
The Global Cancer Atlas Online provides color maps, graphics, and charts that contrast the global
cancer burden. It provides baseline measures by nation for tracking cancer around the world, as well as
data on prevention strategies and cancer risk factors.
Explicitly targeted to “public health professionals”
Interactive cancer rate mapping sites hosted by
individual state registries
Both sexes combined
(default)
Females only
Women of all races (default)
White non-Hispanic
women only
Different agencies providing very different visualizations of same (or
different) data

Involves a particular observation from cancer registry data
 Rapidly changing incidence trend
 Survival disparity among racial/ethnic groups
• Analyze it in detail
• Describe in a scientific manuscript

Publish findings in a peer-reviewed medical journal
• Standard: show visual displays (if any) separately from
communication message (in text)

With journal, disseminate findings as part of press release

Work directly with media to explain findings and relevance
• Visual displays probably based on scientific version
2001-2004 changes
Overall:
-9%
Ages 0-49:
+1%
Ages 70+:
-11%
Ages 50-69:
-12%
ER+:
-15%
ER-:
-2%
No difference by stage
Limited to first primary tumors
2001-2004 changes
Overall:
-9%
Ages 0-49:
+1%
Ages 70+:
-11%
Ages 50-69:
-12%
ER+:
-15%
ER-:
-2%
No difference by stage
Limited to first primary tumors
Ted Kennedy’s glioblastoma
2001-2004 changes
Overall:
-9%
Ages 0-49:
+1%
Ages 70+:
-11%
Ages 50-69:
-12%
ER+:
-15%
ER-:
-2%
No difference by stage
Limited to first primary tumors

Produce annually a roster of local statistics
•
•
•
•

Rates and trends stratified by age, race, year, county
Post on registry website in tabular PDF files
Sometimes an interactive website
Rudimentary (if any) visual displays
Respond to wide array of persons requesting data
• If appropriate, orient them to available online resources
 State resources, State Cancer Profiles, other NCI sites
• Provide customized data
 To a health professional or advocate needing it for
 Health policy, planning, resource allocation
 Setting healthcare priorities
 Targeting specific populations for cancer screening and education
 To the local or national media
• Frequent requests for sub-county level data
 Cancer cluster concern in a neighborhood
 Denominators not generally available at this level
 Statistics often unstable for small populations/rare cancers

Phenomenal job disseminating data to scientific
audiences
• Myriad powerful, interactive internet resources

Little that is appropriate for non-scientists
• Internet resources can be confusing or misleading

Stakes are high
• Data is disseminated widely at local and national levels
 by media after publication in medical journals or press release
 used by policy makers to make hard decisions for prioritization
and planning
 Annual Report to Nation is repackaged and presented by NCI to Congress!

Trying to serve too many masters
• Need different or repackaged data tools for distinct audiences
• “Data smog” due to richness of data

For non-scientific audiences
•
•
•
•
•
•

Do not consider desired response of audience
Not systematic or structured
Do not consider principles of communication science
Do not consider art of visual display
to richness of data
Some statistics (survival, prevalence) easy to misinterpret
Could represent an important gap in public awareness
and understanding of cancer

Training and orientation of cancer surveillance researchers
• Epidemiology and biostatistics
• Little communications training
• Academic scientist orientation
 Think about cancer data in terms of research questions, not communication
goals
• Poor access to experts in communication, design
• Unfamiliarity with design tools other than Microsoft products

Risk aversion
• tendency to avoid sharing data that might be misinterpreted
 Small area data

Times are tough
•
•
registry budgets cut severely in many states
resources diverted away from data use back to data collection
 Has
there ever been any kind of formal needs
assessment for public regarding cancer
statistics?
 Could
existing resources be depackaged and
revised for a lay audience?
 Can
we coordinate information and visual
display across resources: NCI, ACS, CDC?
 How
do we communicate variability or
confidence intervals?
Back page of SEER Landmark
Studies document, 2003